freebsd-skq/contrib/atf/atf-c++/tests.cpp

659 lines
16 KiB
C++
Raw Normal View History

// Copyright (c) 2007 The NetBSD Foundation, Inc.
// All rights reserved.
//
// Redistribution and use in source and binary forms, with or without
// modification, are permitted provided that the following conditions
// are met:
// 1. Redistributions of source code must retain the above copyright
// notice, this list of conditions and the following disclaimer.
// 2. Redistributions in binary form must reproduce the above copyright
// notice, this list of conditions and the following disclaimer in the
// documentation and/or other materials provided with the distribution.
//
// THIS SOFTWARE IS PROVIDED BY THE NETBSD FOUNDATION, INC. AND
// CONTRIBUTORS ``AS IS'' AND ANY EXPRESS OR IMPLIED WARRANTIES,
// INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
// MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.
// IN NO EVENT SHALL THE FOUNDATION OR CONTRIBUTORS BE LIABLE FOR ANY
// DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
// DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE
// GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS
// INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER
// IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR
// OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN
// IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
Import atf-0.21: Released on October 23rd, 2014. * Restored the atf(7) manual page to serve as a reference to all the other manual pages shipped by ATF. * Added the -s flag to atf-sh to support specifying the shell interpreter to be used. * Removed ATF_WORKDIR. The only remaining consumers have been converted to use the standard TMPDIR environment variable. As a benefit, and because Kyua forces the TMPDIR to live within the test case's work directory, any stale files left behind by ATF will be automatically cleaned up. * Documented the environment variables recognized by each component in the relevant manual pages. This information was lost with the atf-config(1) removal. * Added a new "require.diskspace" metadata property to test cases so that they can specify the minimum amount of disk space required for the test to run. * Renamed the atf-{c,c++,sh}-api(3) manual pages to atf-{c,c++,sh}(3) for discoverability purposes. Symbolic links are provided for the time being to still make the old names visible. * Issue #5: Recommend the (expected, actual) idiom for calls to the test macros in the manual pages. * Issue #7: Stopped catching unhandled exceptions in atf-c++ tests. This propagates the crash to the caller, which in turn allows it to obtain proper debugging information. In particular, Kyua should now be able to extract a stacktrace pinpointing the problem. * Issue #8: Fixed atf-c/macros_test:use test failures spotted by the clang that ships with FreeBSD 11.0-CURRENT. * Issue #12: Improved documentation of atf-sh(3) and atf-check(1) by better explaining how they relate to each other. * Issue #14: Stopped setting 'set -e' in atf-sh. This setting was initially added as a way to enable a "strict" mode in the library and to make test cases fail fast when they run unprotected commands. However, doing so in the library is surprising as the responsibility of enabling 'set -e' should be on the user's code. Also, 'set -e' introduces inconsistent behavior on subshells and users do not expect that. * Issue #15: Fixed atf_utils_{fork,wait} to support nested calls. * Issue #16: Fixed test failures (by removing a long-standing hack) on systems that lack \e support in printf(1). * Issue #19: Removed stale references to atf-config and atf-run.
2014-10-30 17:14:07 +00:00
#include "atf-c++/tests.hpp"
#if defined(HAVE_CONFIG_H)
#include "config.h"
#endif
extern "C" {
#include <sys/types.h>
#include <sys/stat.h>
#include <sys/time.h>
#include <sys/wait.h>
#include <signal.h>
#include <unistd.h>
}
#include <algorithm>
#include <cctype>
#include <cerrno>
#include <cstdlib>
#include <cstring>
#include <fstream>
#include <iostream>
#include <map>
#include <memory>
#include <sstream>
#include <stdexcept>
#include <vector>
extern "C" {
#include "atf-c/error.h"
#include "atf-c/tc.h"
#include "atf-c/utils.h"
}
Import atf-0.21: Released on October 23rd, 2014. * Restored the atf(7) manual page to serve as a reference to all the other manual pages shipped by ATF. * Added the -s flag to atf-sh to support specifying the shell interpreter to be used. * Removed ATF_WORKDIR. The only remaining consumers have been converted to use the standard TMPDIR environment variable. As a benefit, and because Kyua forces the TMPDIR to live within the test case's work directory, any stale files left behind by ATF will be automatically cleaned up. * Documented the environment variables recognized by each component in the relevant manual pages. This information was lost with the atf-config(1) removal. * Added a new "require.diskspace" metadata property to test cases so that they can specify the minimum amount of disk space required for the test to run. * Renamed the atf-{c,c++,sh}-api(3) manual pages to atf-{c,c++,sh}(3) for discoverability purposes. Symbolic links are provided for the time being to still make the old names visible. * Issue #5: Recommend the (expected, actual) idiom for calls to the test macros in the manual pages. * Issue #7: Stopped catching unhandled exceptions in atf-c++ tests. This propagates the crash to the caller, which in turn allows it to obtain proper debugging information. In particular, Kyua should now be able to extract a stacktrace pinpointing the problem. * Issue #8: Fixed atf-c/macros_test:use test failures spotted by the clang that ships with FreeBSD 11.0-CURRENT. * Issue #12: Improved documentation of atf-sh(3) and atf-check(1) by better explaining how they relate to each other. * Issue #14: Stopped setting 'set -e' in atf-sh. This setting was initially added as a way to enable a "strict" mode in the library and to make test cases fail fast when they run unprotected commands. However, doing so in the library is surprising as the responsibility of enabling 'set -e' should be on the user's code. Also, 'set -e' introduces inconsistent behavior on subshells and users do not expect that. * Issue #15: Fixed atf_utils_{fork,wait} to support nested calls. * Issue #16: Fixed test failures (by removing a long-standing hack) on systems that lack \e support in printf(1). * Issue #19: Removed stale references to atf-config and atf-run.
2014-10-30 17:14:07 +00:00
#include "atf-c++/detail/application.hpp"
#include "atf-c++/detail/auto_array.hpp"
#include "atf-c++/detail/env.hpp"
#include "atf-c++/detail/exceptions.hpp"
#include "atf-c++/detail/fs.hpp"
#include "atf-c++/detail/sanity.hpp"
#include "atf-c++/detail/text.hpp"
Import atf-0.21: Released on October 23rd, 2014. * Restored the atf(7) manual page to serve as a reference to all the other manual pages shipped by ATF. * Added the -s flag to atf-sh to support specifying the shell interpreter to be used. * Removed ATF_WORKDIR. The only remaining consumers have been converted to use the standard TMPDIR environment variable. As a benefit, and because Kyua forces the TMPDIR to live within the test case's work directory, any stale files left behind by ATF will be automatically cleaned up. * Documented the environment variables recognized by each component in the relevant manual pages. This information was lost with the atf-config(1) removal. * Added a new "require.diskspace" metadata property to test cases so that they can specify the minimum amount of disk space required for the test to run. * Renamed the atf-{c,c++,sh}-api(3) manual pages to atf-{c,c++,sh}(3) for discoverability purposes. Symbolic links are provided for the time being to still make the old names visible. * Issue #5: Recommend the (expected, actual) idiom for calls to the test macros in the manual pages. * Issue #7: Stopped catching unhandled exceptions in atf-c++ tests. This propagates the crash to the caller, which in turn allows it to obtain proper debugging information. In particular, Kyua should now be able to extract a stacktrace pinpointing the problem. * Issue #8: Fixed atf-c/macros_test:use test failures spotted by the clang that ships with FreeBSD 11.0-CURRENT. * Issue #12: Improved documentation of atf-sh(3) and atf-check(1) by better explaining how they relate to each other. * Issue #14: Stopped setting 'set -e' in atf-sh. This setting was initially added as a way to enable a "strict" mode in the library and to make test cases fail fast when they run unprotected commands. However, doing so in the library is surprising as the responsibility of enabling 'set -e' should be on the user's code. Also, 'set -e' introduces inconsistent behavior on subshells and users do not expect that. * Issue #15: Fixed atf_utils_{fork,wait} to support nested calls. * Issue #16: Fixed test failures (by removing a long-standing hack) on systems that lack \e support in printf(1). * Issue #19: Removed stale references to atf-config and atf-run.
2014-10-30 17:14:07 +00:00
#if defined(HAVE_GNU_GETOPT)
# define GETOPT_POSIX "+"
#else
# define GETOPT_POSIX ""
#endif
namespace impl = atf::tests;
namespace detail = atf::tests::detail;
#define IMPL_NAME "atf::tests"
Import atf-0.21: Released on October 23rd, 2014. * Restored the atf(7) manual page to serve as a reference to all the other manual pages shipped by ATF. * Added the -s flag to atf-sh to support specifying the shell interpreter to be used. * Removed ATF_WORKDIR. The only remaining consumers have been converted to use the standard TMPDIR environment variable. As a benefit, and because Kyua forces the TMPDIR to live within the test case's work directory, any stale files left behind by ATF will be automatically cleaned up. * Documented the environment variables recognized by each component in the relevant manual pages. This information was lost with the atf-config(1) removal. * Added a new "require.diskspace" metadata property to test cases so that they can specify the minimum amount of disk space required for the test to run. * Renamed the atf-{c,c++,sh}-api(3) manual pages to atf-{c,c++,sh}(3) for discoverability purposes. Symbolic links are provided for the time being to still make the old names visible. * Issue #5: Recommend the (expected, actual) idiom for calls to the test macros in the manual pages. * Issue #7: Stopped catching unhandled exceptions in atf-c++ tests. This propagates the crash to the caller, which in turn allows it to obtain proper debugging information. In particular, Kyua should now be able to extract a stacktrace pinpointing the problem. * Issue #8: Fixed atf-c/macros_test:use test failures spotted by the clang that ships with FreeBSD 11.0-CURRENT. * Issue #12: Improved documentation of atf-sh(3) and atf-check(1) by better explaining how they relate to each other. * Issue #14: Stopped setting 'set -e' in atf-sh. This setting was initially added as a way to enable a "strict" mode in the library and to make test cases fail fast when they run unprotected commands. However, doing so in the library is surprising as the responsibility of enabling 'set -e' should be on the user's code. Also, 'set -e' introduces inconsistent behavior on subshells and users do not expect that. * Issue #15: Fixed atf_utils_{fork,wait} to support nested calls. * Issue #16: Fixed test failures (by removing a long-standing hack) on systems that lack \e support in printf(1). * Issue #19: Removed stale references to atf-config and atf-run.
2014-10-30 17:14:07 +00:00
using atf::application::usage_error;
// ------------------------------------------------------------------------
// The "atf_tp_writer" class.
// ------------------------------------------------------------------------
detail::atf_tp_writer::atf_tp_writer(std::ostream& os) :
m_os(os),
m_is_first(true)
{
m_os << "Content-Type: application/X-atf-tp; version=\"1\"\n\n";
}
void
detail::atf_tp_writer::start_tc(const std::string& ident)
{
if (!m_is_first)
m_os << "\n";
m_os << "ident: " << ident << "\n";
m_os.flush();
}
void
detail::atf_tp_writer::end_tc(void)
{
if (m_is_first)
m_is_first = false;
}
void
detail::atf_tp_writer::tc_meta_data(const std::string& name,
const std::string& value)
{
PRE(name != "ident");
m_os << name << ": " << value << "\n";
m_os.flush();
}
// ------------------------------------------------------------------------
// Free helper functions.
// ------------------------------------------------------------------------
Import atf-0.21: Released on October 23rd, 2014. * Restored the atf(7) manual page to serve as a reference to all the other manual pages shipped by ATF. * Added the -s flag to atf-sh to support specifying the shell interpreter to be used. * Removed ATF_WORKDIR. The only remaining consumers have been converted to use the standard TMPDIR environment variable. As a benefit, and because Kyua forces the TMPDIR to live within the test case's work directory, any stale files left behind by ATF will be automatically cleaned up. * Documented the environment variables recognized by each component in the relevant manual pages. This information was lost with the atf-config(1) removal. * Added a new "require.diskspace" metadata property to test cases so that they can specify the minimum amount of disk space required for the test to run. * Renamed the atf-{c,c++,sh}-api(3) manual pages to atf-{c,c++,sh}(3) for discoverability purposes. Symbolic links are provided for the time being to still make the old names visible. * Issue #5: Recommend the (expected, actual) idiom for calls to the test macros in the manual pages. * Issue #7: Stopped catching unhandled exceptions in atf-c++ tests. This propagates the crash to the caller, which in turn allows it to obtain proper debugging information. In particular, Kyua should now be able to extract a stacktrace pinpointing the problem. * Issue #8: Fixed atf-c/macros_test:use test failures spotted by the clang that ships with FreeBSD 11.0-CURRENT. * Issue #12: Improved documentation of atf-sh(3) and atf-check(1) by better explaining how they relate to each other. * Issue #14: Stopped setting 'set -e' in atf-sh. This setting was initially added as a way to enable a "strict" mode in the library and to make test cases fail fast when they run unprotected commands. However, doing so in the library is surprising as the responsibility of enabling 'set -e' should be on the user's code. Also, 'set -e' introduces inconsistent behavior on subshells and users do not expect that. * Issue #15: Fixed atf_utils_{fork,wait} to support nested calls. * Issue #16: Fixed test failures (by removing a long-standing hack) on systems that lack \e support in printf(1). * Issue #19: Removed stale references to atf-config and atf-run.
2014-10-30 17:14:07 +00:00
std::string Program_Name;
static void
set_program_name(const char* argv0)
{
const std::string program_name = atf::fs::path(argv0).leaf_name();
// Libtool workaround: if running from within the source tree (binaries
// that are not installed yet), skip the "lt-" prefix added to files in
// the ".libs" directory to show the real (not temporary) name.
if (program_name.substr(0, 3) == "lt-")
Program_Name = program_name.substr(3);
else
Program_Name = program_name;
}
bool
detail::match(const std::string& regexp, const std::string& str)
{
return atf::text::match(str, regexp);
}
// ------------------------------------------------------------------------
// The "tc" class.
// ------------------------------------------------------------------------
static std::map< atf_tc_t*, impl::tc* > wraps;
static std::map< const atf_tc_t*, const impl::tc* > cwraps;
struct impl::tc_impl {
private:
// Non-copyable.
tc_impl(const tc_impl&);
tc_impl& operator=(const tc_impl&);
public:
std::string m_ident;
atf_tc_t m_tc;
bool m_has_cleanup;
tc_impl(const std::string& ident, const bool has_cleanup) :
m_ident(ident),
m_has_cleanup(has_cleanup)
{
}
static void
wrap_head(atf_tc_t *tc)
{
std::map< atf_tc_t*, impl::tc* >::iterator iter = wraps.find(tc);
INV(iter != wraps.end());
(*iter).second->head();
}
static void
wrap_body(const atf_tc_t *tc)
{
std::map< const atf_tc_t*, const impl::tc* >::const_iterator iter =
cwraps.find(tc);
INV(iter != cwraps.end());
Import atf-0.21: Released on October 23rd, 2014. * Restored the atf(7) manual page to serve as a reference to all the other manual pages shipped by ATF. * Added the -s flag to atf-sh to support specifying the shell interpreter to be used. * Removed ATF_WORKDIR. The only remaining consumers have been converted to use the standard TMPDIR environment variable. As a benefit, and because Kyua forces the TMPDIR to live within the test case's work directory, any stale files left behind by ATF will be automatically cleaned up. * Documented the environment variables recognized by each component in the relevant manual pages. This information was lost with the atf-config(1) removal. * Added a new "require.diskspace" metadata property to test cases so that they can specify the minimum amount of disk space required for the test to run. * Renamed the atf-{c,c++,sh}-api(3) manual pages to atf-{c,c++,sh}(3) for discoverability purposes. Symbolic links are provided for the time being to still make the old names visible. * Issue #5: Recommend the (expected, actual) idiom for calls to the test macros in the manual pages. * Issue #7: Stopped catching unhandled exceptions in atf-c++ tests. This propagates the crash to the caller, which in turn allows it to obtain proper debugging information. In particular, Kyua should now be able to extract a stacktrace pinpointing the problem. * Issue #8: Fixed atf-c/macros_test:use test failures spotted by the clang that ships with FreeBSD 11.0-CURRENT. * Issue #12: Improved documentation of atf-sh(3) and atf-check(1) by better explaining how they relate to each other. * Issue #14: Stopped setting 'set -e' in atf-sh. This setting was initially added as a way to enable a "strict" mode in the library and to make test cases fail fast when they run unprotected commands. However, doing so in the library is surprising as the responsibility of enabling 'set -e' should be on the user's code. Also, 'set -e' introduces inconsistent behavior on subshells and users do not expect that. * Issue #15: Fixed atf_utils_{fork,wait} to support nested calls. * Issue #16: Fixed test failures (by removing a long-standing hack) on systems that lack \e support in printf(1). * Issue #19: Removed stale references to atf-config and atf-run.
2014-10-30 17:14:07 +00:00
(*iter).second->body();
}
static void
wrap_cleanup(const atf_tc_t *tc)
{
std::map< const atf_tc_t*, const impl::tc* >::const_iterator iter =
cwraps.find(tc);
INV(iter != cwraps.end());
(*iter).second->cleanup();
}
};
impl::tc::tc(const std::string& ident, const bool has_cleanup) :
pimpl(new tc_impl(ident, has_cleanup))
{
}
impl::tc::~tc(void)
{
cwraps.erase(&pimpl->m_tc);
wraps.erase(&pimpl->m_tc);
atf_tc_fini(&pimpl->m_tc);
}
void
impl::tc::init(const vars_map& config)
{
atf_error_t err;
Import atf-0.17: Experimental version released on February 14th, 2013. * Added the atf_utils_cat_file, atf_utils_compare_file, atf_utils_copy_file, atf_utils_create_file, atf_utils_file_exists, atf_utils_fork, atf_utils_grep_file, atf_utils_grep_string, atf_utils_readline, atf_utils_redirect and atf_utils_wait utility functions to atf-c-api. Documented the already-public atf_utils_free_charpp function. * Added the cat_file, compare_file, copy_file, create_file, file_exists, fork, grep_collection, grep_file, grep_string, redirect and wait functions to the atf::utils namespace of atf-c++-api. These are wrappers around the same functions added to the atf-c-api library. * Added the ATF_CHECK_MATCH, ATF_CHECK_MATCH_MSG, ATF_REQUIRE_MATCH and ATF_REQUIRE_MATCH_MSG macros to atf-c to simplify the validation of a string against a regular expression. * Miscellaneous fixes for manpage typos and compilation problems with clang. * Added caching of the results of those configure tests that rely on executing a test program. This should help crossbuild systems by providing a mechanism to pre-specify what the results should be. * X-NetBSD-PR bin/45690: Make atf-report convert any non-printable characters to a plain-text representation (matching their corresponding hexadecimal entities) in XML output files. This is to prevent the output of test cases from breaking xsltproc later. Note that this import, compared to the one for 0.16, brings in all the files that are part of the release. This is to follow the Subversion Primer guidelines, which mention that all files should be imported first and only dropped when merging into contrib/atf/. Approved by: rpaulo (mentor)
2013-11-15 21:28:06 +00:00
auto_array< const char * > array(new const char*[(config.size() * 2) + 1]);
const char **ptr = array.get();
for (vars_map::const_iterator iter = config.begin();
iter != config.end(); iter++) {
*ptr = (*iter).first.c_str();
*(ptr + 1) = (*iter).second.c_str();
ptr += 2;
}
*ptr = NULL;
wraps[&pimpl->m_tc] = this;
cwraps[&pimpl->m_tc] = this;
err = atf_tc_init(&pimpl->m_tc, pimpl->m_ident.c_str(), pimpl->wrap_head,
pimpl->wrap_body, pimpl->m_has_cleanup ? pimpl->wrap_cleanup : NULL,
array.get());
if (atf_is_error(err))
throw_atf_error(err);
}
bool
impl::tc::has_config_var(const std::string& var)
const
{
return atf_tc_has_config_var(&pimpl->m_tc, var.c_str());
}
bool
impl::tc::has_md_var(const std::string& var)
const
{
return atf_tc_has_md_var(&pimpl->m_tc, var.c_str());
}
const std::string
impl::tc::get_config_var(const std::string& var)
const
{
return atf_tc_get_config_var(&pimpl->m_tc, var.c_str());
}
const std::string
impl::tc::get_config_var(const std::string& var, const std::string& defval)
const
{
return atf_tc_get_config_var_wd(&pimpl->m_tc, var.c_str(), defval.c_str());
}
const std::string
impl::tc::get_md_var(const std::string& var)
const
{
return atf_tc_get_md_var(&pimpl->m_tc, var.c_str());
}
const impl::vars_map
impl::tc::get_md_vars(void)
const
{
vars_map vars;
char **array = atf_tc_get_md_vars(&pimpl->m_tc);
try {
char **ptr;
for (ptr = array; *ptr != NULL; ptr += 2)
vars[*ptr] = *(ptr + 1);
} catch (...) {
atf_utils_free_charpp(array);
throw;
}
return vars;
}
void
impl::tc::set_md_var(const std::string& var, const std::string& val)
{
atf_error_t err = atf_tc_set_md_var(&pimpl->m_tc, var.c_str(), val.c_str());
if (atf_is_error(err))
throw_atf_error(err);
}
void
impl::tc::run(const std::string& resfile)
const
{
atf_error_t err = atf_tc_run(&pimpl->m_tc, resfile.c_str());
if (atf_is_error(err))
throw_atf_error(err);
}
void
impl::tc::run_cleanup(void)
const
{
atf_error_t err = atf_tc_cleanup(&pimpl->m_tc);
if (atf_is_error(err))
throw_atf_error(err);
}
void
impl::tc::head(void)
{
}
void
impl::tc::cleanup(void)
const
{
}
void
impl::tc::require_prog(const std::string& prog)
const
{
atf_tc_require_prog(prog.c_str());
}
void
impl::tc::pass(void)
{
atf_tc_pass();
}
void
impl::tc::fail(const std::string& reason)
{
atf_tc_fail("%s", reason.c_str());
}
void
impl::tc::fail_nonfatal(const std::string& reason)
{
atf_tc_fail_nonfatal("%s", reason.c_str());
}
void
impl::tc::skip(const std::string& reason)
{
atf_tc_skip("%s", reason.c_str());
}
void
impl::tc::check_errno(const char* file, const int line, const int exp_errno,
const char* expr_str, const bool result)
{
atf_tc_check_errno(file, line, exp_errno, expr_str, result);
}
void
impl::tc::require_errno(const char* file, const int line, const int exp_errno,
const char* expr_str, const bool result)
{
atf_tc_require_errno(file, line, exp_errno, expr_str, result);
}
void
impl::tc::expect_pass(void)
{
atf_tc_expect_pass();
}
void
impl::tc::expect_fail(const std::string& reason)
{
atf_tc_expect_fail("%s", reason.c_str());
}
void
impl::tc::expect_exit(const int exitcode, const std::string& reason)
{
atf_tc_expect_exit(exitcode, "%s", reason.c_str());
}
void
impl::tc::expect_signal(const int signo, const std::string& reason)
{
atf_tc_expect_signal(signo, "%s", reason.c_str());
}
void
impl::tc::expect_death(const std::string& reason)
{
atf_tc_expect_death("%s", reason.c_str());
}
void
impl::tc::expect_timeout(const std::string& reason)
{
atf_tc_expect_timeout("%s", reason.c_str());
}
// ------------------------------------------------------------------------
Import atf-0.21: Released on October 23rd, 2014. * Restored the atf(7) manual page to serve as a reference to all the other manual pages shipped by ATF. * Added the -s flag to atf-sh to support specifying the shell interpreter to be used. * Removed ATF_WORKDIR. The only remaining consumers have been converted to use the standard TMPDIR environment variable. As a benefit, and because Kyua forces the TMPDIR to live within the test case's work directory, any stale files left behind by ATF will be automatically cleaned up. * Documented the environment variables recognized by each component in the relevant manual pages. This information was lost with the atf-config(1) removal. * Added a new "require.diskspace" metadata property to test cases so that they can specify the minimum amount of disk space required for the test to run. * Renamed the atf-{c,c++,sh}-api(3) manual pages to atf-{c,c++,sh}(3) for discoverability purposes. Symbolic links are provided for the time being to still make the old names visible. * Issue #5: Recommend the (expected, actual) idiom for calls to the test macros in the manual pages. * Issue #7: Stopped catching unhandled exceptions in atf-c++ tests. This propagates the crash to the caller, which in turn allows it to obtain proper debugging information. In particular, Kyua should now be able to extract a stacktrace pinpointing the problem. * Issue #8: Fixed atf-c/macros_test:use test failures spotted by the clang that ships with FreeBSD 11.0-CURRENT. * Issue #12: Improved documentation of atf-sh(3) and atf-check(1) by better explaining how they relate to each other. * Issue #14: Stopped setting 'set -e' in atf-sh. This setting was initially added as a way to enable a "strict" mode in the library and to make test cases fail fast when they run unprotected commands. However, doing so in the library is surprising as the responsibility of enabling 'set -e' should be on the user's code. Also, 'set -e' introduces inconsistent behavior on subshells and users do not expect that. * Issue #15: Fixed atf_utils_{fork,wait} to support nested calls. * Issue #16: Fixed test failures (by removing a long-standing hack) on systems that lack \e support in printf(1). * Issue #19: Removed stale references to atf-config and atf-run.
2014-10-30 17:14:07 +00:00
// Test program main code.
// ------------------------------------------------------------------------
Import atf-0.21: Released on October 23rd, 2014. * Restored the atf(7) manual page to serve as a reference to all the other manual pages shipped by ATF. * Added the -s flag to atf-sh to support specifying the shell interpreter to be used. * Removed ATF_WORKDIR. The only remaining consumers have been converted to use the standard TMPDIR environment variable. As a benefit, and because Kyua forces the TMPDIR to live within the test case's work directory, any stale files left behind by ATF will be automatically cleaned up. * Documented the environment variables recognized by each component in the relevant manual pages. This information was lost with the atf-config(1) removal. * Added a new "require.diskspace" metadata property to test cases so that they can specify the minimum amount of disk space required for the test to run. * Renamed the atf-{c,c++,sh}-api(3) manual pages to atf-{c,c++,sh}(3) for discoverability purposes. Symbolic links are provided for the time being to still make the old names visible. * Issue #5: Recommend the (expected, actual) idiom for calls to the test macros in the manual pages. * Issue #7: Stopped catching unhandled exceptions in atf-c++ tests. This propagates the crash to the caller, which in turn allows it to obtain proper debugging information. In particular, Kyua should now be able to extract a stacktrace pinpointing the problem. * Issue #8: Fixed atf-c/macros_test:use test failures spotted by the clang that ships with FreeBSD 11.0-CURRENT. * Issue #12: Improved documentation of atf-sh(3) and atf-check(1) by better explaining how they relate to each other. * Issue #14: Stopped setting 'set -e' in atf-sh. This setting was initially added as a way to enable a "strict" mode in the library and to make test cases fail fast when they run unprotected commands. However, doing so in the library is surprising as the responsibility of enabling 'set -e' should be on the user's code. Also, 'set -e' introduces inconsistent behavior on subshells and users do not expect that. * Issue #15: Fixed atf_utils_{fork,wait} to support nested calls. * Issue #16: Fixed test failures (by removing a long-standing hack) on systems that lack \e support in printf(1). * Issue #19: Removed stale references to atf-config and atf-run.
2014-10-30 17:14:07 +00:00
namespace {
Import atf-0.21: Released on October 23rd, 2014. * Restored the atf(7) manual page to serve as a reference to all the other manual pages shipped by ATF. * Added the -s flag to atf-sh to support specifying the shell interpreter to be used. * Removed ATF_WORKDIR. The only remaining consumers have been converted to use the standard TMPDIR environment variable. As a benefit, and because Kyua forces the TMPDIR to live within the test case's work directory, any stale files left behind by ATF will be automatically cleaned up. * Documented the environment variables recognized by each component in the relevant manual pages. This information was lost with the atf-config(1) removal. * Added a new "require.diskspace" metadata property to test cases so that they can specify the minimum amount of disk space required for the test to run. * Renamed the atf-{c,c++,sh}-api(3) manual pages to atf-{c,c++,sh}(3) for discoverability purposes. Symbolic links are provided for the time being to still make the old names visible. * Issue #5: Recommend the (expected, actual) idiom for calls to the test macros in the manual pages. * Issue #7: Stopped catching unhandled exceptions in atf-c++ tests. This propagates the crash to the caller, which in turn allows it to obtain proper debugging information. In particular, Kyua should now be able to extract a stacktrace pinpointing the problem. * Issue #8: Fixed atf-c/macros_test:use test failures spotted by the clang that ships with FreeBSD 11.0-CURRENT. * Issue #12: Improved documentation of atf-sh(3) and atf-check(1) by better explaining how they relate to each other. * Issue #14: Stopped setting 'set -e' in atf-sh. This setting was initially added as a way to enable a "strict" mode in the library and to make test cases fail fast when they run unprotected commands. However, doing so in the library is surprising as the responsibility of enabling 'set -e' should be on the user's code. Also, 'set -e' introduces inconsistent behavior on subshells and users do not expect that. * Issue #15: Fixed atf_utils_{fork,wait} to support nested calls. * Issue #16: Fixed test failures (by removing a long-standing hack) on systems that lack \e support in printf(1). * Issue #19: Removed stale references to atf-config and atf-run.
2014-10-30 17:14:07 +00:00
typedef std::vector< impl::tc * > tc_vector;
Import atf-0.21: Released on October 23rd, 2014. * Restored the atf(7) manual page to serve as a reference to all the other manual pages shipped by ATF. * Added the -s flag to atf-sh to support specifying the shell interpreter to be used. * Removed ATF_WORKDIR. The only remaining consumers have been converted to use the standard TMPDIR environment variable. As a benefit, and because Kyua forces the TMPDIR to live within the test case's work directory, any stale files left behind by ATF will be automatically cleaned up. * Documented the environment variables recognized by each component in the relevant manual pages. This information was lost with the atf-config(1) removal. * Added a new "require.diskspace" metadata property to test cases so that they can specify the minimum amount of disk space required for the test to run. * Renamed the atf-{c,c++,sh}-api(3) manual pages to atf-{c,c++,sh}(3) for discoverability purposes. Symbolic links are provided for the time being to still make the old names visible. * Issue #5: Recommend the (expected, actual) idiom for calls to the test macros in the manual pages. * Issue #7: Stopped catching unhandled exceptions in atf-c++ tests. This propagates the crash to the caller, which in turn allows it to obtain proper debugging information. In particular, Kyua should now be able to extract a stacktrace pinpointing the problem. * Issue #8: Fixed atf-c/macros_test:use test failures spotted by the clang that ships with FreeBSD 11.0-CURRENT. * Issue #12: Improved documentation of atf-sh(3) and atf-check(1) by better explaining how they relate to each other. * Issue #14: Stopped setting 'set -e' in atf-sh. This setting was initially added as a way to enable a "strict" mode in the library and to make test cases fail fast when they run unprotected commands. However, doing so in the library is surprising as the responsibility of enabling 'set -e' should be on the user's code. Also, 'set -e' introduces inconsistent behavior on subshells and users do not expect that. * Issue #15: Fixed atf_utils_{fork,wait} to support nested calls. * Issue #16: Fixed test failures (by removing a long-standing hack) on systems that lack \e support in printf(1). * Issue #19: Removed stale references to atf-config and atf-run.
2014-10-30 17:14:07 +00:00
enum tc_part { BODY, CLEANUP };
Import atf-0.21: Released on October 23rd, 2014. * Restored the atf(7) manual page to serve as a reference to all the other manual pages shipped by ATF. * Added the -s flag to atf-sh to support specifying the shell interpreter to be used. * Removed ATF_WORKDIR. The only remaining consumers have been converted to use the standard TMPDIR environment variable. As a benefit, and because Kyua forces the TMPDIR to live within the test case's work directory, any stale files left behind by ATF will be automatically cleaned up. * Documented the environment variables recognized by each component in the relevant manual pages. This information was lost with the atf-config(1) removal. * Added a new "require.diskspace" metadata property to test cases so that they can specify the minimum amount of disk space required for the test to run. * Renamed the atf-{c,c++,sh}-api(3) manual pages to atf-{c,c++,sh}(3) for discoverability purposes. Symbolic links are provided for the time being to still make the old names visible. * Issue #5: Recommend the (expected, actual) idiom for calls to the test macros in the manual pages. * Issue #7: Stopped catching unhandled exceptions in atf-c++ tests. This propagates the crash to the caller, which in turn allows it to obtain proper debugging information. In particular, Kyua should now be able to extract a stacktrace pinpointing the problem. * Issue #8: Fixed atf-c/macros_test:use test failures spotted by the clang that ships with FreeBSD 11.0-CURRENT. * Issue #12: Improved documentation of atf-sh(3) and atf-check(1) by better explaining how they relate to each other. * Issue #14: Stopped setting 'set -e' in atf-sh. This setting was initially added as a way to enable a "strict" mode in the library and to make test cases fail fast when they run unprotected commands. However, doing so in the library is surprising as the responsibility of enabling 'set -e' should be on the user's code. Also, 'set -e' introduces inconsistent behavior on subshells and users do not expect that. * Issue #15: Fixed atf_utils_{fork,wait} to support nested calls. * Issue #16: Fixed test failures (by removing a long-standing hack) on systems that lack \e support in printf(1). * Issue #19: Removed stale references to atf-config and atf-run.
2014-10-30 17:14:07 +00:00
static void
parse_vflag(const std::string& str, atf::tests::vars_map& vars)
{
if (str.empty())
throw std::runtime_error("-v requires a non-empty argument");
std::vector< std::string > ws = atf::text::split(str, "=");
if (ws.size() == 1 && str[str.length() - 1] == '=') {
Import atf-0.21: Released on October 23rd, 2014. * Restored the atf(7) manual page to serve as a reference to all the other manual pages shipped by ATF. * Added the -s flag to atf-sh to support specifying the shell interpreter to be used. * Removed ATF_WORKDIR. The only remaining consumers have been converted to use the standard TMPDIR environment variable. As a benefit, and because Kyua forces the TMPDIR to live within the test case's work directory, any stale files left behind by ATF will be automatically cleaned up. * Documented the environment variables recognized by each component in the relevant manual pages. This information was lost with the atf-config(1) removal. * Added a new "require.diskspace" metadata property to test cases so that they can specify the minimum amount of disk space required for the test to run. * Renamed the atf-{c,c++,sh}-api(3) manual pages to atf-{c,c++,sh}(3) for discoverability purposes. Symbolic links are provided for the time being to still make the old names visible. * Issue #5: Recommend the (expected, actual) idiom for calls to the test macros in the manual pages. * Issue #7: Stopped catching unhandled exceptions in atf-c++ tests. This propagates the crash to the caller, which in turn allows it to obtain proper debugging information. In particular, Kyua should now be able to extract a stacktrace pinpointing the problem. * Issue #8: Fixed atf-c/macros_test:use test failures spotted by the clang that ships with FreeBSD 11.0-CURRENT. * Issue #12: Improved documentation of atf-sh(3) and atf-check(1) by better explaining how they relate to each other. * Issue #14: Stopped setting 'set -e' in atf-sh. This setting was initially added as a way to enable a "strict" mode in the library and to make test cases fail fast when they run unprotected commands. However, doing so in the library is surprising as the responsibility of enabling 'set -e' should be on the user's code. Also, 'set -e' introduces inconsistent behavior on subshells and users do not expect that. * Issue #15: Fixed atf_utils_{fork,wait} to support nested calls. * Issue #16: Fixed test failures (by removing a long-standing hack) on systems that lack \e support in printf(1). * Issue #19: Removed stale references to atf-config and atf-run.
2014-10-30 17:14:07 +00:00
vars[ws[0]] = "";
} else {
if (ws.size() != 2)
throw std::runtime_error("-v requires an argument of the form "
"var=value");
Import atf-0.21: Released on October 23rd, 2014. * Restored the atf(7) manual page to serve as a reference to all the other manual pages shipped by ATF. * Added the -s flag to atf-sh to support specifying the shell interpreter to be used. * Removed ATF_WORKDIR. The only remaining consumers have been converted to use the standard TMPDIR environment variable. As a benefit, and because Kyua forces the TMPDIR to live within the test case's work directory, any stale files left behind by ATF will be automatically cleaned up. * Documented the environment variables recognized by each component in the relevant manual pages. This information was lost with the atf-config(1) removal. * Added a new "require.diskspace" metadata property to test cases so that they can specify the minimum amount of disk space required for the test to run. * Renamed the atf-{c,c++,sh}-api(3) manual pages to atf-{c,c++,sh}(3) for discoverability purposes. Symbolic links are provided for the time being to still make the old names visible. * Issue #5: Recommend the (expected, actual) idiom for calls to the test macros in the manual pages. * Issue #7: Stopped catching unhandled exceptions in atf-c++ tests. This propagates the crash to the caller, which in turn allows it to obtain proper debugging information. In particular, Kyua should now be able to extract a stacktrace pinpointing the problem. * Issue #8: Fixed atf-c/macros_test:use test failures spotted by the clang that ships with FreeBSD 11.0-CURRENT. * Issue #12: Improved documentation of atf-sh(3) and atf-check(1) by better explaining how they relate to each other. * Issue #14: Stopped setting 'set -e' in atf-sh. This setting was initially added as a way to enable a "strict" mode in the library and to make test cases fail fast when they run unprotected commands. However, doing so in the library is surprising as the responsibility of enabling 'set -e' should be on the user's code. Also, 'set -e' introduces inconsistent behavior on subshells and users do not expect that. * Issue #15: Fixed atf_utils_{fork,wait} to support nested calls. * Issue #16: Fixed test failures (by removing a long-standing hack) on systems that lack \e support in printf(1). * Issue #19: Removed stale references to atf-config and atf-run.
2014-10-30 17:14:07 +00:00
vars[ws[0]] = ws[1];
}
}
Import atf-0.21: Released on October 23rd, 2014. * Restored the atf(7) manual page to serve as a reference to all the other manual pages shipped by ATF. * Added the -s flag to atf-sh to support specifying the shell interpreter to be used. * Removed ATF_WORKDIR. The only remaining consumers have been converted to use the standard TMPDIR environment variable. As a benefit, and because Kyua forces the TMPDIR to live within the test case's work directory, any stale files left behind by ATF will be automatically cleaned up. * Documented the environment variables recognized by each component in the relevant manual pages. This information was lost with the atf-config(1) removal. * Added a new "require.diskspace" metadata property to test cases so that they can specify the minimum amount of disk space required for the test to run. * Renamed the atf-{c,c++,sh}-api(3) manual pages to atf-{c,c++,sh}(3) for discoverability purposes. Symbolic links are provided for the time being to still make the old names visible. * Issue #5: Recommend the (expected, actual) idiom for calls to the test macros in the manual pages. * Issue #7: Stopped catching unhandled exceptions in atf-c++ tests. This propagates the crash to the caller, which in turn allows it to obtain proper debugging information. In particular, Kyua should now be able to extract a stacktrace pinpointing the problem. * Issue #8: Fixed atf-c/macros_test:use test failures spotted by the clang that ships with FreeBSD 11.0-CURRENT. * Issue #12: Improved documentation of atf-sh(3) and atf-check(1) by better explaining how they relate to each other. * Issue #14: Stopped setting 'set -e' in atf-sh. This setting was initially added as a way to enable a "strict" mode in the library and to make test cases fail fast when they run unprotected commands. However, doing so in the library is surprising as the responsibility of enabling 'set -e' should be on the user's code. Also, 'set -e' introduces inconsistent behavior on subshells and users do not expect that. * Issue #15: Fixed atf_utils_{fork,wait} to support nested calls. * Issue #16: Fixed test failures (by removing a long-standing hack) on systems that lack \e support in printf(1). * Issue #19: Removed stale references to atf-config and atf-run.
2014-10-30 17:14:07 +00:00
static atf::fs::path
handle_srcdir(const char* argv0, const std::string& srcdir_arg)
{
Import atf-0.21: Released on October 23rd, 2014. * Restored the atf(7) manual page to serve as a reference to all the other manual pages shipped by ATF. * Added the -s flag to atf-sh to support specifying the shell interpreter to be used. * Removed ATF_WORKDIR. The only remaining consumers have been converted to use the standard TMPDIR environment variable. As a benefit, and because Kyua forces the TMPDIR to live within the test case's work directory, any stale files left behind by ATF will be automatically cleaned up. * Documented the environment variables recognized by each component in the relevant manual pages. This information was lost with the atf-config(1) removal. * Added a new "require.diskspace" metadata property to test cases so that they can specify the minimum amount of disk space required for the test to run. * Renamed the atf-{c,c++,sh}-api(3) manual pages to atf-{c,c++,sh}(3) for discoverability purposes. Symbolic links are provided for the time being to still make the old names visible. * Issue #5: Recommend the (expected, actual) idiom for calls to the test macros in the manual pages. * Issue #7: Stopped catching unhandled exceptions in atf-c++ tests. This propagates the crash to the caller, which in turn allows it to obtain proper debugging information. In particular, Kyua should now be able to extract a stacktrace pinpointing the problem. * Issue #8: Fixed atf-c/macros_test:use test failures spotted by the clang that ships with FreeBSD 11.0-CURRENT. * Issue #12: Improved documentation of atf-sh(3) and atf-check(1) by better explaining how they relate to each other. * Issue #14: Stopped setting 'set -e' in atf-sh. This setting was initially added as a way to enable a "strict" mode in the library and to make test cases fail fast when they run unprotected commands. However, doing so in the library is surprising as the responsibility of enabling 'set -e' should be on the user's code. Also, 'set -e' introduces inconsistent behavior on subshells and users do not expect that. * Issue #15: Fixed atf_utils_{fork,wait} to support nested calls. * Issue #16: Fixed test failures (by removing a long-standing hack) on systems that lack \e support in printf(1). * Issue #19: Removed stale references to atf-config and atf-run.
2014-10-30 17:14:07 +00:00
atf::fs::path srcdir(".");
if (srcdir_arg.empty()) {
srcdir = atf::fs::path(argv0).branch_path();
if (srcdir.leaf_name() == ".libs")
srcdir = srcdir.branch_path();
} else
Import atf-0.21: Released on October 23rd, 2014. * Restored the atf(7) manual page to serve as a reference to all the other manual pages shipped by ATF. * Added the -s flag to atf-sh to support specifying the shell interpreter to be used. * Removed ATF_WORKDIR. The only remaining consumers have been converted to use the standard TMPDIR environment variable. As a benefit, and because Kyua forces the TMPDIR to live within the test case's work directory, any stale files left behind by ATF will be automatically cleaned up. * Documented the environment variables recognized by each component in the relevant manual pages. This information was lost with the atf-config(1) removal. * Added a new "require.diskspace" metadata property to test cases so that they can specify the minimum amount of disk space required for the test to run. * Renamed the atf-{c,c++,sh}-api(3) manual pages to atf-{c,c++,sh}(3) for discoverability purposes. Symbolic links are provided for the time being to still make the old names visible. * Issue #5: Recommend the (expected, actual) idiom for calls to the test macros in the manual pages. * Issue #7: Stopped catching unhandled exceptions in atf-c++ tests. This propagates the crash to the caller, which in turn allows it to obtain proper debugging information. In particular, Kyua should now be able to extract a stacktrace pinpointing the problem. * Issue #8: Fixed atf-c/macros_test:use test failures spotted by the clang that ships with FreeBSD 11.0-CURRENT. * Issue #12: Improved documentation of atf-sh(3) and atf-check(1) by better explaining how they relate to each other. * Issue #14: Stopped setting 'set -e' in atf-sh. This setting was initially added as a way to enable a "strict" mode in the library and to make test cases fail fast when they run unprotected commands. However, doing so in the library is surprising as the responsibility of enabling 'set -e' should be on the user's code. Also, 'set -e' introduces inconsistent behavior on subshells and users do not expect that. * Issue #15: Fixed atf_utils_{fork,wait} to support nested calls. * Issue #16: Fixed test failures (by removing a long-standing hack) on systems that lack \e support in printf(1). * Issue #19: Removed stale references to atf-config and atf-run.
2014-10-30 17:14:07 +00:00
srcdir = atf::fs::path(srcdir_arg);
Import atf-0.21: Released on October 23rd, 2014. * Restored the atf(7) manual page to serve as a reference to all the other manual pages shipped by ATF. * Added the -s flag to atf-sh to support specifying the shell interpreter to be used. * Removed ATF_WORKDIR. The only remaining consumers have been converted to use the standard TMPDIR environment variable. As a benefit, and because Kyua forces the TMPDIR to live within the test case's work directory, any stale files left behind by ATF will be automatically cleaned up. * Documented the environment variables recognized by each component in the relevant manual pages. This information was lost with the atf-config(1) removal. * Added a new "require.diskspace" metadata property to test cases so that they can specify the minimum amount of disk space required for the test to run. * Renamed the atf-{c,c++,sh}-api(3) manual pages to atf-{c,c++,sh}(3) for discoverability purposes. Symbolic links are provided for the time being to still make the old names visible. * Issue #5: Recommend the (expected, actual) idiom for calls to the test macros in the manual pages. * Issue #7: Stopped catching unhandled exceptions in atf-c++ tests. This propagates the crash to the caller, which in turn allows it to obtain proper debugging information. In particular, Kyua should now be able to extract a stacktrace pinpointing the problem. * Issue #8: Fixed atf-c/macros_test:use test failures spotted by the clang that ships with FreeBSD 11.0-CURRENT. * Issue #12: Improved documentation of atf-sh(3) and atf-check(1) by better explaining how they relate to each other. * Issue #14: Stopped setting 'set -e' in atf-sh. This setting was initially added as a way to enable a "strict" mode in the library and to make test cases fail fast when they run unprotected commands. However, doing so in the library is surprising as the responsibility of enabling 'set -e' should be on the user's code. Also, 'set -e' introduces inconsistent behavior on subshells and users do not expect that. * Issue #15: Fixed atf_utils_{fork,wait} to support nested calls. * Issue #16: Fixed test failures (by removing a long-standing hack) on systems that lack \e support in printf(1). * Issue #19: Removed stale references to atf-config and atf-run.
2014-10-30 17:14:07 +00:00
if (!atf::fs::exists(srcdir / Program_Name))
throw usage_error("Cannot find the test program in the source "
"directory `%s'", srcdir.c_str());
Import atf-0.21: Released on October 23rd, 2014. * Restored the atf(7) manual page to serve as a reference to all the other manual pages shipped by ATF. * Added the -s flag to atf-sh to support specifying the shell interpreter to be used. * Removed ATF_WORKDIR. The only remaining consumers have been converted to use the standard TMPDIR environment variable. As a benefit, and because Kyua forces the TMPDIR to live within the test case's work directory, any stale files left behind by ATF will be automatically cleaned up. * Documented the environment variables recognized by each component in the relevant manual pages. This information was lost with the atf-config(1) removal. * Added a new "require.diskspace" metadata property to test cases so that they can specify the minimum amount of disk space required for the test to run. * Renamed the atf-{c,c++,sh}-api(3) manual pages to atf-{c,c++,sh}(3) for discoverability purposes. Symbolic links are provided for the time being to still make the old names visible. * Issue #5: Recommend the (expected, actual) idiom for calls to the test macros in the manual pages. * Issue #7: Stopped catching unhandled exceptions in atf-c++ tests. This propagates the crash to the caller, which in turn allows it to obtain proper debugging information. In particular, Kyua should now be able to extract a stacktrace pinpointing the problem. * Issue #8: Fixed atf-c/macros_test:use test failures spotted by the clang that ships with FreeBSD 11.0-CURRENT. * Issue #12: Improved documentation of atf-sh(3) and atf-check(1) by better explaining how they relate to each other. * Issue #14: Stopped setting 'set -e' in atf-sh. This setting was initially added as a way to enable a "strict" mode in the library and to make test cases fail fast when they run unprotected commands. However, doing so in the library is surprising as the responsibility of enabling 'set -e' should be on the user's code. Also, 'set -e' introduces inconsistent behavior on subshells and users do not expect that. * Issue #15: Fixed atf_utils_{fork,wait} to support nested calls. * Issue #16: Fixed test failures (by removing a long-standing hack) on systems that lack \e support in printf(1). * Issue #19: Removed stale references to atf-config and atf-run.
2014-10-30 17:14:07 +00:00
if (!srcdir.is_absolute())
srcdir = srcdir.to_absolute();
Import atf-0.21: Released on October 23rd, 2014. * Restored the atf(7) manual page to serve as a reference to all the other manual pages shipped by ATF. * Added the -s flag to atf-sh to support specifying the shell interpreter to be used. * Removed ATF_WORKDIR. The only remaining consumers have been converted to use the standard TMPDIR environment variable. As a benefit, and because Kyua forces the TMPDIR to live within the test case's work directory, any stale files left behind by ATF will be automatically cleaned up. * Documented the environment variables recognized by each component in the relevant manual pages. This information was lost with the atf-config(1) removal. * Added a new "require.diskspace" metadata property to test cases so that they can specify the minimum amount of disk space required for the test to run. * Renamed the atf-{c,c++,sh}-api(3) manual pages to atf-{c,c++,sh}(3) for discoverability purposes. Symbolic links are provided for the time being to still make the old names visible. * Issue #5: Recommend the (expected, actual) idiom for calls to the test macros in the manual pages. * Issue #7: Stopped catching unhandled exceptions in atf-c++ tests. This propagates the crash to the caller, which in turn allows it to obtain proper debugging information. In particular, Kyua should now be able to extract a stacktrace pinpointing the problem. * Issue #8: Fixed atf-c/macros_test:use test failures spotted by the clang that ships with FreeBSD 11.0-CURRENT. * Issue #12: Improved documentation of atf-sh(3) and atf-check(1) by better explaining how they relate to each other. * Issue #14: Stopped setting 'set -e' in atf-sh. This setting was initially added as a way to enable a "strict" mode in the library and to make test cases fail fast when they run unprotected commands. However, doing so in the library is surprising as the responsibility of enabling 'set -e' should be on the user's code. Also, 'set -e' introduces inconsistent behavior on subshells and users do not expect that. * Issue #15: Fixed atf_utils_{fork,wait} to support nested calls. * Issue #16: Fixed test failures (by removing a long-standing hack) on systems that lack \e support in printf(1). * Issue #19: Removed stale references to atf-config and atf-run.
2014-10-30 17:14:07 +00:00
return srcdir;
}
Import atf-0.21: Released on October 23rd, 2014. * Restored the atf(7) manual page to serve as a reference to all the other manual pages shipped by ATF. * Added the -s flag to atf-sh to support specifying the shell interpreter to be used. * Removed ATF_WORKDIR. The only remaining consumers have been converted to use the standard TMPDIR environment variable. As a benefit, and because Kyua forces the TMPDIR to live within the test case's work directory, any stale files left behind by ATF will be automatically cleaned up. * Documented the environment variables recognized by each component in the relevant manual pages. This information was lost with the atf-config(1) removal. * Added a new "require.diskspace" metadata property to test cases so that they can specify the minimum amount of disk space required for the test to run. * Renamed the atf-{c,c++,sh}-api(3) manual pages to atf-{c,c++,sh}(3) for discoverability purposes. Symbolic links are provided for the time being to still make the old names visible. * Issue #5: Recommend the (expected, actual) idiom for calls to the test macros in the manual pages. * Issue #7: Stopped catching unhandled exceptions in atf-c++ tests. This propagates the crash to the caller, which in turn allows it to obtain proper debugging information. In particular, Kyua should now be able to extract a stacktrace pinpointing the problem. * Issue #8: Fixed atf-c/macros_test:use test failures spotted by the clang that ships with FreeBSD 11.0-CURRENT. * Issue #12: Improved documentation of atf-sh(3) and atf-check(1) by better explaining how they relate to each other. * Issue #14: Stopped setting 'set -e' in atf-sh. This setting was initially added as a way to enable a "strict" mode in the library and to make test cases fail fast when they run unprotected commands. However, doing so in the library is surprising as the responsibility of enabling 'set -e' should be on the user's code. Also, 'set -e' introduces inconsistent behavior on subshells and users do not expect that. * Issue #15: Fixed atf_utils_{fork,wait} to support nested calls. * Issue #16: Fixed test failures (by removing a long-standing hack) on systems that lack \e support in printf(1). * Issue #19: Removed stale references to atf-config and atf-run.
2014-10-30 17:14:07 +00:00
static void
init_tcs(void (*add_tcs)(tc_vector&), tc_vector& tcs,
const atf::tests::vars_map& vars)
{
Import atf-0.21: Released on October 23rd, 2014. * Restored the atf(7) manual page to serve as a reference to all the other manual pages shipped by ATF. * Added the -s flag to atf-sh to support specifying the shell interpreter to be used. * Removed ATF_WORKDIR. The only remaining consumers have been converted to use the standard TMPDIR environment variable. As a benefit, and because Kyua forces the TMPDIR to live within the test case's work directory, any stale files left behind by ATF will be automatically cleaned up. * Documented the environment variables recognized by each component in the relevant manual pages. This information was lost with the atf-config(1) removal. * Added a new "require.diskspace" metadata property to test cases so that they can specify the minimum amount of disk space required for the test to run. * Renamed the atf-{c,c++,sh}-api(3) manual pages to atf-{c,c++,sh}(3) for discoverability purposes. Symbolic links are provided for the time being to still make the old names visible. * Issue #5: Recommend the (expected, actual) idiom for calls to the test macros in the manual pages. * Issue #7: Stopped catching unhandled exceptions in atf-c++ tests. This propagates the crash to the caller, which in turn allows it to obtain proper debugging information. In particular, Kyua should now be able to extract a stacktrace pinpointing the problem. * Issue #8: Fixed atf-c/macros_test:use test failures spotted by the clang that ships with FreeBSD 11.0-CURRENT. * Issue #12: Improved documentation of atf-sh(3) and atf-check(1) by better explaining how they relate to each other. * Issue #14: Stopped setting 'set -e' in atf-sh. This setting was initially added as a way to enable a "strict" mode in the library and to make test cases fail fast when they run unprotected commands. However, doing so in the library is surprising as the responsibility of enabling 'set -e' should be on the user's code. Also, 'set -e' introduces inconsistent behavior on subshells and users do not expect that. * Issue #15: Fixed atf_utils_{fork,wait} to support nested calls. * Issue #16: Fixed test failures (by removing a long-standing hack) on systems that lack \e support in printf(1). * Issue #19: Removed stale references to atf-config and atf-run.
2014-10-30 17:14:07 +00:00
add_tcs(tcs);
for (tc_vector::iterator iter = tcs.begin(); iter != tcs.end(); iter++) {
impl::tc* tc = *iter;
Import atf-0.21: Released on October 23rd, 2014. * Restored the atf(7) manual page to serve as a reference to all the other manual pages shipped by ATF. * Added the -s flag to atf-sh to support specifying the shell interpreter to be used. * Removed ATF_WORKDIR. The only remaining consumers have been converted to use the standard TMPDIR environment variable. As a benefit, and because Kyua forces the TMPDIR to live within the test case's work directory, any stale files left behind by ATF will be automatically cleaned up. * Documented the environment variables recognized by each component in the relevant manual pages. This information was lost with the atf-config(1) removal. * Added a new "require.diskspace" metadata property to test cases so that they can specify the minimum amount of disk space required for the test to run. * Renamed the atf-{c,c++,sh}-api(3) manual pages to atf-{c,c++,sh}(3) for discoverability purposes. Symbolic links are provided for the time being to still make the old names visible. * Issue #5: Recommend the (expected, actual) idiom for calls to the test macros in the manual pages. * Issue #7: Stopped catching unhandled exceptions in atf-c++ tests. This propagates the crash to the caller, which in turn allows it to obtain proper debugging information. In particular, Kyua should now be able to extract a stacktrace pinpointing the problem. * Issue #8: Fixed atf-c/macros_test:use test failures spotted by the clang that ships with FreeBSD 11.0-CURRENT. * Issue #12: Improved documentation of atf-sh(3) and atf-check(1) by better explaining how they relate to each other. * Issue #14: Stopped setting 'set -e' in atf-sh. This setting was initially added as a way to enable a "strict" mode in the library and to make test cases fail fast when they run unprotected commands. However, doing so in the library is surprising as the responsibility of enabling 'set -e' should be on the user's code. Also, 'set -e' introduces inconsistent behavior on subshells and users do not expect that. * Issue #15: Fixed atf_utils_{fork,wait} to support nested calls. * Issue #16: Fixed test failures (by removing a long-standing hack) on systems that lack \e support in printf(1). * Issue #19: Removed stale references to atf-config and atf-run.
2014-10-30 17:14:07 +00:00
tc->init(vars);
}
}
Import atf-0.21: Released on October 23rd, 2014. * Restored the atf(7) manual page to serve as a reference to all the other manual pages shipped by ATF. * Added the -s flag to atf-sh to support specifying the shell interpreter to be used. * Removed ATF_WORKDIR. The only remaining consumers have been converted to use the standard TMPDIR environment variable. As a benefit, and because Kyua forces the TMPDIR to live within the test case's work directory, any stale files left behind by ATF will be automatically cleaned up. * Documented the environment variables recognized by each component in the relevant manual pages. This information was lost with the atf-config(1) removal. * Added a new "require.diskspace" metadata property to test cases so that they can specify the minimum amount of disk space required for the test to run. * Renamed the atf-{c,c++,sh}-api(3) manual pages to atf-{c,c++,sh}(3) for discoverability purposes. Symbolic links are provided for the time being to still make the old names visible. * Issue #5: Recommend the (expected, actual) idiom for calls to the test macros in the manual pages. * Issue #7: Stopped catching unhandled exceptions in atf-c++ tests. This propagates the crash to the caller, which in turn allows it to obtain proper debugging information. In particular, Kyua should now be able to extract a stacktrace pinpointing the problem. * Issue #8: Fixed atf-c/macros_test:use test failures spotted by the clang that ships with FreeBSD 11.0-CURRENT. * Issue #12: Improved documentation of atf-sh(3) and atf-check(1) by better explaining how they relate to each other. * Issue #14: Stopped setting 'set -e' in atf-sh. This setting was initially added as a way to enable a "strict" mode in the library and to make test cases fail fast when they run unprotected commands. However, doing so in the library is surprising as the responsibility of enabling 'set -e' should be on the user's code. Also, 'set -e' introduces inconsistent behavior on subshells and users do not expect that. * Issue #15: Fixed atf_utils_{fork,wait} to support nested calls. * Issue #16: Fixed test failures (by removing a long-standing hack) on systems that lack \e support in printf(1). * Issue #19: Removed stale references to atf-config and atf-run.
2014-10-30 17:14:07 +00:00
static int
list_tcs(const tc_vector& tcs)
{
detail::atf_tp_writer writer(std::cout);
for (tc_vector::const_iterator iter = tcs.begin();
iter != tcs.end(); iter++) {
const impl::vars_map vars = (*iter)->get_md_vars();
{
impl::vars_map::const_iterator iter2 = vars.find("ident");
INV(iter2 != vars.end());
writer.start_tc((*iter2).second);
}
for (impl::vars_map::const_iterator iter2 = vars.begin();
iter2 != vars.end(); iter2++) {
const std::string& key = (*iter2).first;
if (key != "ident")
writer.tc_meta_data(key, (*iter2).second);
}
writer.end_tc();
}
Import atf-0.21: Released on October 23rd, 2014. * Restored the atf(7) manual page to serve as a reference to all the other manual pages shipped by ATF. * Added the -s flag to atf-sh to support specifying the shell interpreter to be used. * Removed ATF_WORKDIR. The only remaining consumers have been converted to use the standard TMPDIR environment variable. As a benefit, and because Kyua forces the TMPDIR to live within the test case's work directory, any stale files left behind by ATF will be automatically cleaned up. * Documented the environment variables recognized by each component in the relevant manual pages. This information was lost with the atf-config(1) removal. * Added a new "require.diskspace" metadata property to test cases so that they can specify the minimum amount of disk space required for the test to run. * Renamed the atf-{c,c++,sh}-api(3) manual pages to atf-{c,c++,sh}(3) for discoverability purposes. Symbolic links are provided for the time being to still make the old names visible. * Issue #5: Recommend the (expected, actual) idiom for calls to the test macros in the manual pages. * Issue #7: Stopped catching unhandled exceptions in atf-c++ tests. This propagates the crash to the caller, which in turn allows it to obtain proper debugging information. In particular, Kyua should now be able to extract a stacktrace pinpointing the problem. * Issue #8: Fixed atf-c/macros_test:use test failures spotted by the clang that ships with FreeBSD 11.0-CURRENT. * Issue #12: Improved documentation of atf-sh(3) and atf-check(1) by better explaining how they relate to each other. * Issue #14: Stopped setting 'set -e' in atf-sh. This setting was initially added as a way to enable a "strict" mode in the library and to make test cases fail fast when they run unprotected commands. However, doing so in the library is surprising as the responsibility of enabling 'set -e' should be on the user's code. Also, 'set -e' introduces inconsistent behavior on subshells and users do not expect that. * Issue #15: Fixed atf_utils_{fork,wait} to support nested calls. * Issue #16: Fixed test failures (by removing a long-standing hack) on systems that lack \e support in printf(1). * Issue #19: Removed stale references to atf-config and atf-run.
2014-10-30 17:14:07 +00:00
return EXIT_SUCCESS;
}
Import atf-0.21: Released on October 23rd, 2014. * Restored the atf(7) manual page to serve as a reference to all the other manual pages shipped by ATF. * Added the -s flag to atf-sh to support specifying the shell interpreter to be used. * Removed ATF_WORKDIR. The only remaining consumers have been converted to use the standard TMPDIR environment variable. As a benefit, and because Kyua forces the TMPDIR to live within the test case's work directory, any stale files left behind by ATF will be automatically cleaned up. * Documented the environment variables recognized by each component in the relevant manual pages. This information was lost with the atf-config(1) removal. * Added a new "require.diskspace" metadata property to test cases so that they can specify the minimum amount of disk space required for the test to run. * Renamed the atf-{c,c++,sh}-api(3) manual pages to atf-{c,c++,sh}(3) for discoverability purposes. Symbolic links are provided for the time being to still make the old names visible. * Issue #5: Recommend the (expected, actual) idiom for calls to the test macros in the manual pages. * Issue #7: Stopped catching unhandled exceptions in atf-c++ tests. This propagates the crash to the caller, which in turn allows it to obtain proper debugging information. In particular, Kyua should now be able to extract a stacktrace pinpointing the problem. * Issue #8: Fixed atf-c/macros_test:use test failures spotted by the clang that ships with FreeBSD 11.0-CURRENT. * Issue #12: Improved documentation of atf-sh(3) and atf-check(1) by better explaining how they relate to each other. * Issue #14: Stopped setting 'set -e' in atf-sh. This setting was initially added as a way to enable a "strict" mode in the library and to make test cases fail fast when they run unprotected commands. However, doing so in the library is surprising as the responsibility of enabling 'set -e' should be on the user's code. Also, 'set -e' introduces inconsistent behavior on subshells and users do not expect that. * Issue #15: Fixed atf_utils_{fork,wait} to support nested calls. * Issue #16: Fixed test failures (by removing a long-standing hack) on systems that lack \e support in printf(1). * Issue #19: Removed stale references to atf-config and atf-run.
2014-10-30 17:14:07 +00:00
static impl::tc*
find_tc(tc_vector tcs, const std::string& name)
{
std::vector< std::string > ids;
for (tc_vector::iterator iter = tcs.begin();
iter != tcs.end(); iter++) {
impl::tc* tc = *iter;
if (tc->get_md_var("ident") == name)
return tc;
}
Import atf-0.21: Released on October 23rd, 2014. * Restored the atf(7) manual page to serve as a reference to all the other manual pages shipped by ATF. * Added the -s flag to atf-sh to support specifying the shell interpreter to be used. * Removed ATF_WORKDIR. The only remaining consumers have been converted to use the standard TMPDIR environment variable. As a benefit, and because Kyua forces the TMPDIR to live within the test case's work directory, any stale files left behind by ATF will be automatically cleaned up. * Documented the environment variables recognized by each component in the relevant manual pages. This information was lost with the atf-config(1) removal. * Added a new "require.diskspace" metadata property to test cases so that they can specify the minimum amount of disk space required for the test to run. * Renamed the atf-{c,c++,sh}-api(3) manual pages to atf-{c,c++,sh}(3) for discoverability purposes. Symbolic links are provided for the time being to still make the old names visible. * Issue #5: Recommend the (expected, actual) idiom for calls to the test macros in the manual pages. * Issue #7: Stopped catching unhandled exceptions in atf-c++ tests. This propagates the crash to the caller, which in turn allows it to obtain proper debugging information. In particular, Kyua should now be able to extract a stacktrace pinpointing the problem. * Issue #8: Fixed atf-c/macros_test:use test failures spotted by the clang that ships with FreeBSD 11.0-CURRENT. * Issue #12: Improved documentation of atf-sh(3) and atf-check(1) by better explaining how they relate to each other. * Issue #14: Stopped setting 'set -e' in atf-sh. This setting was initially added as a way to enable a "strict" mode in the library and to make test cases fail fast when they run unprotected commands. However, doing so in the library is surprising as the responsibility of enabling 'set -e' should be on the user's code. Also, 'set -e' introduces inconsistent behavior on subshells and users do not expect that. * Issue #15: Fixed atf_utils_{fork,wait} to support nested calls. * Issue #16: Fixed test failures (by removing a long-standing hack) on systems that lack \e support in printf(1). * Issue #19: Removed stale references to atf-config and atf-run.
2014-10-30 17:14:07 +00:00
throw usage_error("Unknown test case `%s'", name.c_str());
}
Import atf-0.21: Released on October 23rd, 2014. * Restored the atf(7) manual page to serve as a reference to all the other manual pages shipped by ATF. * Added the -s flag to atf-sh to support specifying the shell interpreter to be used. * Removed ATF_WORKDIR. The only remaining consumers have been converted to use the standard TMPDIR environment variable. As a benefit, and because Kyua forces the TMPDIR to live within the test case's work directory, any stale files left behind by ATF will be automatically cleaned up. * Documented the environment variables recognized by each component in the relevant manual pages. This information was lost with the atf-config(1) removal. * Added a new "require.diskspace" metadata property to test cases so that they can specify the minimum amount of disk space required for the test to run. * Renamed the atf-{c,c++,sh}-api(3) manual pages to atf-{c,c++,sh}(3) for discoverability purposes. Symbolic links are provided for the time being to still make the old names visible. * Issue #5: Recommend the (expected, actual) idiom for calls to the test macros in the manual pages. * Issue #7: Stopped catching unhandled exceptions in atf-c++ tests. This propagates the crash to the caller, which in turn allows it to obtain proper debugging information. In particular, Kyua should now be able to extract a stacktrace pinpointing the problem. * Issue #8: Fixed atf-c/macros_test:use test failures spotted by the clang that ships with FreeBSD 11.0-CURRENT. * Issue #12: Improved documentation of atf-sh(3) and atf-check(1) by better explaining how they relate to each other. * Issue #14: Stopped setting 'set -e' in atf-sh. This setting was initially added as a way to enable a "strict" mode in the library and to make test cases fail fast when they run unprotected commands. However, doing so in the library is surprising as the responsibility of enabling 'set -e' should be on the user's code. Also, 'set -e' introduces inconsistent behavior on subshells and users do not expect that. * Issue #15: Fixed atf_utils_{fork,wait} to support nested calls. * Issue #16: Fixed test failures (by removing a long-standing hack) on systems that lack \e support in printf(1). * Issue #19: Removed stale references to atf-config and atf-run.
2014-10-30 17:14:07 +00:00
static std::pair< std::string, tc_part >
process_tcarg(const std::string& tcarg)
{
const std::string::size_type pos = tcarg.find(':');
if (pos == std::string::npos) {
return std::make_pair(tcarg, BODY);
} else {
const std::string tcname = tcarg.substr(0, pos);
const std::string partname = tcarg.substr(pos + 1);
if (partname == "body")
return std::make_pair(tcname, BODY);
else if (partname == "cleanup")
return std::make_pair(tcname, CLEANUP);
else {
throw usage_error("Invalid test case part `%s'", partname.c_str());
}
}
}
Import atf-0.21: Released on October 23rd, 2014. * Restored the atf(7) manual page to serve as a reference to all the other manual pages shipped by ATF. * Added the -s flag to atf-sh to support specifying the shell interpreter to be used. * Removed ATF_WORKDIR. The only remaining consumers have been converted to use the standard TMPDIR environment variable. As a benefit, and because Kyua forces the TMPDIR to live within the test case's work directory, any stale files left behind by ATF will be automatically cleaned up. * Documented the environment variables recognized by each component in the relevant manual pages. This information was lost with the atf-config(1) removal. * Added a new "require.diskspace" metadata property to test cases so that they can specify the minimum amount of disk space required for the test to run. * Renamed the atf-{c,c++,sh}-api(3) manual pages to atf-{c,c++,sh}(3) for discoverability purposes. Symbolic links are provided for the time being to still make the old names visible. * Issue #5: Recommend the (expected, actual) idiom for calls to the test macros in the manual pages. * Issue #7: Stopped catching unhandled exceptions in atf-c++ tests. This propagates the crash to the caller, which in turn allows it to obtain proper debugging information. In particular, Kyua should now be able to extract a stacktrace pinpointing the problem. * Issue #8: Fixed atf-c/macros_test:use test failures spotted by the clang that ships with FreeBSD 11.0-CURRENT. * Issue #12: Improved documentation of atf-sh(3) and atf-check(1) by better explaining how they relate to each other. * Issue #14: Stopped setting 'set -e' in atf-sh. This setting was initially added as a way to enable a "strict" mode in the library and to make test cases fail fast when they run unprotected commands. However, doing so in the library is surprising as the responsibility of enabling 'set -e' should be on the user's code. Also, 'set -e' introduces inconsistent behavior on subshells and users do not expect that. * Issue #15: Fixed atf_utils_{fork,wait} to support nested calls. * Issue #16: Fixed test failures (by removing a long-standing hack) on systems that lack \e support in printf(1). * Issue #19: Removed stale references to atf-config and atf-run.
2014-10-30 17:14:07 +00:00
static int
run_tc(tc_vector& tcs, const std::string& tcarg, const atf::fs::path& resfile)
{
const std::pair< std::string, tc_part > fields = process_tcarg(tcarg);
Import atf-0.21: Released on October 23rd, 2014. * Restored the atf(7) manual page to serve as a reference to all the other manual pages shipped by ATF. * Added the -s flag to atf-sh to support specifying the shell interpreter to be used. * Removed ATF_WORKDIR. The only remaining consumers have been converted to use the standard TMPDIR environment variable. As a benefit, and because Kyua forces the TMPDIR to live within the test case's work directory, any stale files left behind by ATF will be automatically cleaned up. * Documented the environment variables recognized by each component in the relevant manual pages. This information was lost with the atf-config(1) removal. * Added a new "require.diskspace" metadata property to test cases so that they can specify the minimum amount of disk space required for the test to run. * Renamed the atf-{c,c++,sh}-api(3) manual pages to atf-{c,c++,sh}(3) for discoverability purposes. Symbolic links are provided for the time being to still make the old names visible. * Issue #5: Recommend the (expected, actual) idiom for calls to the test macros in the manual pages. * Issue #7: Stopped catching unhandled exceptions in atf-c++ tests. This propagates the crash to the caller, which in turn allows it to obtain proper debugging information. In particular, Kyua should now be able to extract a stacktrace pinpointing the problem. * Issue #8: Fixed atf-c/macros_test:use test failures spotted by the clang that ships with FreeBSD 11.0-CURRENT. * Issue #12: Improved documentation of atf-sh(3) and atf-check(1) by better explaining how they relate to each other. * Issue #14: Stopped setting 'set -e' in atf-sh. This setting was initially added as a way to enable a "strict" mode in the library and to make test cases fail fast when they run unprotected commands. However, doing so in the library is surprising as the responsibility of enabling 'set -e' should be on the user's code. Also, 'set -e' introduces inconsistent behavior on subshells and users do not expect that. * Issue #15: Fixed atf_utils_{fork,wait} to support nested calls. * Issue #16: Fixed test failures (by removing a long-standing hack) on systems that lack \e support in printf(1). * Issue #19: Removed stale references to atf-config and atf-run.
2014-10-30 17:14:07 +00:00
impl::tc* tc = find_tc(tcs, fields.first);
if (!atf::env::has("__RUNNING_INSIDE_ATF_RUN") || atf::env::get(
"__RUNNING_INSIDE_ATF_RUN") != "internal-yes-value")
{
Import atf-0.21: Released on October 23rd, 2014. * Restored the atf(7) manual page to serve as a reference to all the other manual pages shipped by ATF. * Added the -s flag to atf-sh to support specifying the shell interpreter to be used. * Removed ATF_WORKDIR. The only remaining consumers have been converted to use the standard TMPDIR environment variable. As a benefit, and because Kyua forces the TMPDIR to live within the test case's work directory, any stale files left behind by ATF will be automatically cleaned up. * Documented the environment variables recognized by each component in the relevant manual pages. This information was lost with the atf-config(1) removal. * Added a new "require.diskspace" metadata property to test cases so that they can specify the minimum amount of disk space required for the test to run. * Renamed the atf-{c,c++,sh}-api(3) manual pages to atf-{c,c++,sh}(3) for discoverability purposes. Symbolic links are provided for the time being to still make the old names visible. * Issue #5: Recommend the (expected, actual) idiom for calls to the test macros in the manual pages. * Issue #7: Stopped catching unhandled exceptions in atf-c++ tests. This propagates the crash to the caller, which in turn allows it to obtain proper debugging information. In particular, Kyua should now be able to extract a stacktrace pinpointing the problem. * Issue #8: Fixed atf-c/macros_test:use test failures spotted by the clang that ships with FreeBSD 11.0-CURRENT. * Issue #12: Improved documentation of atf-sh(3) and atf-check(1) by better explaining how they relate to each other. * Issue #14: Stopped setting 'set -e' in atf-sh. This setting was initially added as a way to enable a "strict" mode in the library and to make test cases fail fast when they run unprotected commands. However, doing so in the library is surprising as the responsibility of enabling 'set -e' should be on the user's code. Also, 'set -e' introduces inconsistent behavior on subshells and users do not expect that. * Issue #15: Fixed atf_utils_{fork,wait} to support nested calls. * Issue #16: Fixed test failures (by removing a long-standing hack) on systems that lack \e support in printf(1). * Issue #19: Removed stale references to atf-config and atf-run.
2014-10-30 17:14:07 +00:00
std::cerr << Program_Name << ": WARNING: Running test cases outside "
"of kyua(1) is unsupported\n";
std::cerr << Program_Name << ": WARNING: No isolation nor timeout "
"control is being applied; you may get unexpected failures; see "
"atf-test-case(4)\n";
}
Import atf-0.21: Released on October 23rd, 2014. * Restored the atf(7) manual page to serve as a reference to all the other manual pages shipped by ATF. * Added the -s flag to atf-sh to support specifying the shell interpreter to be used. * Removed ATF_WORKDIR. The only remaining consumers have been converted to use the standard TMPDIR environment variable. As a benefit, and because Kyua forces the TMPDIR to live within the test case's work directory, any stale files left behind by ATF will be automatically cleaned up. * Documented the environment variables recognized by each component in the relevant manual pages. This information was lost with the atf-config(1) removal. * Added a new "require.diskspace" metadata property to test cases so that they can specify the minimum amount of disk space required for the test to run. * Renamed the atf-{c,c++,sh}-api(3) manual pages to atf-{c,c++,sh}(3) for discoverability purposes. Symbolic links are provided for the time being to still make the old names visible. * Issue #5: Recommend the (expected, actual) idiom for calls to the test macros in the manual pages. * Issue #7: Stopped catching unhandled exceptions in atf-c++ tests. This propagates the crash to the caller, which in turn allows it to obtain proper debugging information. In particular, Kyua should now be able to extract a stacktrace pinpointing the problem. * Issue #8: Fixed atf-c/macros_test:use test failures spotted by the clang that ships with FreeBSD 11.0-CURRENT. * Issue #12: Improved documentation of atf-sh(3) and atf-check(1) by better explaining how they relate to each other. * Issue #14: Stopped setting 'set -e' in atf-sh. This setting was initially added as a way to enable a "strict" mode in the library and to make test cases fail fast when they run unprotected commands. However, doing so in the library is surprising as the responsibility of enabling 'set -e' should be on the user's code. Also, 'set -e' introduces inconsistent behavior on subshells and users do not expect that. * Issue #15: Fixed atf_utils_{fork,wait} to support nested calls. * Issue #16: Fixed test failures (by removing a long-standing hack) on systems that lack \e support in printf(1). * Issue #19: Removed stale references to atf-config and atf-run.
2014-10-30 17:14:07 +00:00
switch (fields.second) {
case BODY:
tc->run(resfile.str());
break;
case CLEANUP:
tc->run_cleanup();
break;
default:
UNREACHABLE;
}
return EXIT_SUCCESS;
}
static int
safe_main(int argc, char** argv, void (*add_tcs)(tc_vector&))
{
const char* argv0 = argv[0];
bool lflag = false;
atf::fs::path resfile("/dev/stdout");
std::string srcdir_arg;
atf::tests::vars_map vars;
int ch;
int old_opterr;
old_opterr = opterr;
::opterr = 0;
while ((ch = ::getopt(argc, argv, GETOPT_POSIX ":lr:s:v:")) != -1) {
switch (ch) {
case 'l':
lflag = true;
break;
case 'r':
resfile = atf::fs::path(::optarg);
break;
Import atf-0.21: Released on October 23rd, 2014. * Restored the atf(7) manual page to serve as a reference to all the other manual pages shipped by ATF. * Added the -s flag to atf-sh to support specifying the shell interpreter to be used. * Removed ATF_WORKDIR. The only remaining consumers have been converted to use the standard TMPDIR environment variable. As a benefit, and because Kyua forces the TMPDIR to live within the test case's work directory, any stale files left behind by ATF will be automatically cleaned up. * Documented the environment variables recognized by each component in the relevant manual pages. This information was lost with the atf-config(1) removal. * Added a new "require.diskspace" metadata property to test cases so that they can specify the minimum amount of disk space required for the test to run. * Renamed the atf-{c,c++,sh}-api(3) manual pages to atf-{c,c++,sh}(3) for discoverability purposes. Symbolic links are provided for the time being to still make the old names visible. * Issue #5: Recommend the (expected, actual) idiom for calls to the test macros in the manual pages. * Issue #7: Stopped catching unhandled exceptions in atf-c++ tests. This propagates the crash to the caller, which in turn allows it to obtain proper debugging information. In particular, Kyua should now be able to extract a stacktrace pinpointing the problem. * Issue #8: Fixed atf-c/macros_test:use test failures spotted by the clang that ships with FreeBSD 11.0-CURRENT. * Issue #12: Improved documentation of atf-sh(3) and atf-check(1) by better explaining how they relate to each other. * Issue #14: Stopped setting 'set -e' in atf-sh. This setting was initially added as a way to enable a "strict" mode in the library and to make test cases fail fast when they run unprotected commands. However, doing so in the library is surprising as the responsibility of enabling 'set -e' should be on the user's code. Also, 'set -e' introduces inconsistent behavior on subshells and users do not expect that. * Issue #15: Fixed atf_utils_{fork,wait} to support nested calls. * Issue #16: Fixed test failures (by removing a long-standing hack) on systems that lack \e support in printf(1). * Issue #19: Removed stale references to atf-config and atf-run.
2014-10-30 17:14:07 +00:00
case 's':
srcdir_arg = ::optarg;
break;
case 'v':
parse_vflag(::optarg, vars);
break;
Import atf-0.21: Released on October 23rd, 2014. * Restored the atf(7) manual page to serve as a reference to all the other manual pages shipped by ATF. * Added the -s flag to atf-sh to support specifying the shell interpreter to be used. * Removed ATF_WORKDIR. The only remaining consumers have been converted to use the standard TMPDIR environment variable. As a benefit, and because Kyua forces the TMPDIR to live within the test case's work directory, any stale files left behind by ATF will be automatically cleaned up. * Documented the environment variables recognized by each component in the relevant manual pages. This information was lost with the atf-config(1) removal. * Added a new "require.diskspace" metadata property to test cases so that they can specify the minimum amount of disk space required for the test to run. * Renamed the atf-{c,c++,sh}-api(3) manual pages to atf-{c,c++,sh}(3) for discoverability purposes. Symbolic links are provided for the time being to still make the old names visible. * Issue #5: Recommend the (expected, actual) idiom for calls to the test macros in the manual pages. * Issue #7: Stopped catching unhandled exceptions in atf-c++ tests. This propagates the crash to the caller, which in turn allows it to obtain proper debugging information. In particular, Kyua should now be able to extract a stacktrace pinpointing the problem. * Issue #8: Fixed atf-c/macros_test:use test failures spotted by the clang that ships with FreeBSD 11.0-CURRENT. * Issue #12: Improved documentation of atf-sh(3) and atf-check(1) by better explaining how they relate to each other. * Issue #14: Stopped setting 'set -e' in atf-sh. This setting was initially added as a way to enable a "strict" mode in the library and to make test cases fail fast when they run unprotected commands. However, doing so in the library is surprising as the responsibility of enabling 'set -e' should be on the user's code. Also, 'set -e' introduces inconsistent behavior on subshells and users do not expect that. * Issue #15: Fixed atf_utils_{fork,wait} to support nested calls. * Issue #16: Fixed test failures (by removing a long-standing hack) on systems that lack \e support in printf(1). * Issue #19: Removed stale references to atf-config and atf-run.
2014-10-30 17:14:07 +00:00
case ':':
throw usage_error("Option -%c requires an argument.", ::optopt);
break;
case '?':
default:
Import atf-0.21: Released on October 23rd, 2014. * Restored the atf(7) manual page to serve as a reference to all the other manual pages shipped by ATF. * Added the -s flag to atf-sh to support specifying the shell interpreter to be used. * Removed ATF_WORKDIR. The only remaining consumers have been converted to use the standard TMPDIR environment variable. As a benefit, and because Kyua forces the TMPDIR to live within the test case's work directory, any stale files left behind by ATF will be automatically cleaned up. * Documented the environment variables recognized by each component in the relevant manual pages. This information was lost with the atf-config(1) removal. * Added a new "require.diskspace" metadata property to test cases so that they can specify the minimum amount of disk space required for the test to run. * Renamed the atf-{c,c++,sh}-api(3) manual pages to atf-{c,c++,sh}(3) for discoverability purposes. Symbolic links are provided for the time being to still make the old names visible. * Issue #5: Recommend the (expected, actual) idiom for calls to the test macros in the manual pages. * Issue #7: Stopped catching unhandled exceptions in atf-c++ tests. This propagates the crash to the caller, which in turn allows it to obtain proper debugging information. In particular, Kyua should now be able to extract a stacktrace pinpointing the problem. * Issue #8: Fixed atf-c/macros_test:use test failures spotted by the clang that ships with FreeBSD 11.0-CURRENT. * Issue #12: Improved documentation of atf-sh(3) and atf-check(1) by better explaining how they relate to each other. * Issue #14: Stopped setting 'set -e' in atf-sh. This setting was initially added as a way to enable a "strict" mode in the library and to make test cases fail fast when they run unprotected commands. However, doing so in the library is surprising as the responsibility of enabling 'set -e' should be on the user's code. Also, 'set -e' introduces inconsistent behavior on subshells and users do not expect that. * Issue #15: Fixed atf_utils_{fork,wait} to support nested calls. * Issue #16: Fixed test failures (by removing a long-standing hack) on systems that lack \e support in printf(1). * Issue #19: Removed stale references to atf-config and atf-run.
2014-10-30 17:14:07 +00:00
throw usage_error("Unknown option -%c.", ::optopt);
}
}
Import atf-0.21: Released on October 23rd, 2014. * Restored the atf(7) manual page to serve as a reference to all the other manual pages shipped by ATF. * Added the -s flag to atf-sh to support specifying the shell interpreter to be used. * Removed ATF_WORKDIR. The only remaining consumers have been converted to use the standard TMPDIR environment variable. As a benefit, and because Kyua forces the TMPDIR to live within the test case's work directory, any stale files left behind by ATF will be automatically cleaned up. * Documented the environment variables recognized by each component in the relevant manual pages. This information was lost with the atf-config(1) removal. * Added a new "require.diskspace" metadata property to test cases so that they can specify the minimum amount of disk space required for the test to run. * Renamed the atf-{c,c++,sh}-api(3) manual pages to atf-{c,c++,sh}(3) for discoverability purposes. Symbolic links are provided for the time being to still make the old names visible. * Issue #5: Recommend the (expected, actual) idiom for calls to the test macros in the manual pages. * Issue #7: Stopped catching unhandled exceptions in atf-c++ tests. This propagates the crash to the caller, which in turn allows it to obtain proper debugging information. In particular, Kyua should now be able to extract a stacktrace pinpointing the problem. * Issue #8: Fixed atf-c/macros_test:use test failures spotted by the clang that ships with FreeBSD 11.0-CURRENT. * Issue #12: Improved documentation of atf-sh(3) and atf-check(1) by better explaining how they relate to each other. * Issue #14: Stopped setting 'set -e' in atf-sh. This setting was initially added as a way to enable a "strict" mode in the library and to make test cases fail fast when they run unprotected commands. However, doing so in the library is surprising as the responsibility of enabling 'set -e' should be on the user's code. Also, 'set -e' introduces inconsistent behavior on subshells and users do not expect that. * Issue #15: Fixed atf_utils_{fork,wait} to support nested calls. * Issue #16: Fixed test failures (by removing a long-standing hack) on systems that lack \e support in printf(1). * Issue #19: Removed stale references to atf-config and atf-run.
2014-10-30 17:14:07 +00:00
argc -= optind;
argv += optind;
Import atf-0.21: Released on October 23rd, 2014. * Restored the atf(7) manual page to serve as a reference to all the other manual pages shipped by ATF. * Added the -s flag to atf-sh to support specifying the shell interpreter to be used. * Removed ATF_WORKDIR. The only remaining consumers have been converted to use the standard TMPDIR environment variable. As a benefit, and because Kyua forces the TMPDIR to live within the test case's work directory, any stale files left behind by ATF will be automatically cleaned up. * Documented the environment variables recognized by each component in the relevant manual pages. This information was lost with the atf-config(1) removal. * Added a new "require.diskspace" metadata property to test cases so that they can specify the minimum amount of disk space required for the test to run. * Renamed the atf-{c,c++,sh}-api(3) manual pages to atf-{c,c++,sh}(3) for discoverability purposes. Symbolic links are provided for the time being to still make the old names visible. * Issue #5: Recommend the (expected, actual) idiom for calls to the test macros in the manual pages. * Issue #7: Stopped catching unhandled exceptions in atf-c++ tests. This propagates the crash to the caller, which in turn allows it to obtain proper debugging information. In particular, Kyua should now be able to extract a stacktrace pinpointing the problem. * Issue #8: Fixed atf-c/macros_test:use test failures spotted by the clang that ships with FreeBSD 11.0-CURRENT. * Issue #12: Improved documentation of atf-sh(3) and atf-check(1) by better explaining how they relate to each other. * Issue #14: Stopped setting 'set -e' in atf-sh. This setting was initially added as a way to enable a "strict" mode in the library and to make test cases fail fast when they run unprotected commands. However, doing so in the library is surprising as the responsibility of enabling 'set -e' should be on the user's code. Also, 'set -e' introduces inconsistent behavior on subshells and users do not expect that. * Issue #15: Fixed atf_utils_{fork,wait} to support nested calls. * Issue #16: Fixed test failures (by removing a long-standing hack) on systems that lack \e support in printf(1). * Issue #19: Removed stale references to atf-config and atf-run.
2014-10-30 17:14:07 +00:00
// Clear getopt state just in case the test wants to use it.
::opterr = old_opterr;
::optind = 1;
#if defined(HAVE_OPTRESET)
::optreset = 1;
#endif
Import atf-0.21: Released on October 23rd, 2014. * Restored the atf(7) manual page to serve as a reference to all the other manual pages shipped by ATF. * Added the -s flag to atf-sh to support specifying the shell interpreter to be used. * Removed ATF_WORKDIR. The only remaining consumers have been converted to use the standard TMPDIR environment variable. As a benefit, and because Kyua forces the TMPDIR to live within the test case's work directory, any stale files left behind by ATF will be automatically cleaned up. * Documented the environment variables recognized by each component in the relevant manual pages. This information was lost with the atf-config(1) removal. * Added a new "require.diskspace" metadata property to test cases so that they can specify the minimum amount of disk space required for the test to run. * Renamed the atf-{c,c++,sh}-api(3) manual pages to atf-{c,c++,sh}(3) for discoverability purposes. Symbolic links are provided for the time being to still make the old names visible. * Issue #5: Recommend the (expected, actual) idiom for calls to the test macros in the manual pages. * Issue #7: Stopped catching unhandled exceptions in atf-c++ tests. This propagates the crash to the caller, which in turn allows it to obtain proper debugging information. In particular, Kyua should now be able to extract a stacktrace pinpointing the problem. * Issue #8: Fixed atf-c/macros_test:use test failures spotted by the clang that ships with FreeBSD 11.0-CURRENT. * Issue #12: Improved documentation of atf-sh(3) and atf-check(1) by better explaining how they relate to each other. * Issue #14: Stopped setting 'set -e' in atf-sh. This setting was initially added as a way to enable a "strict" mode in the library and to make test cases fail fast when they run unprotected commands. However, doing so in the library is surprising as the responsibility of enabling 'set -e' should be on the user's code. Also, 'set -e' introduces inconsistent behavior on subshells and users do not expect that. * Issue #15: Fixed atf_utils_{fork,wait} to support nested calls. * Issue #16: Fixed test failures (by removing a long-standing hack) on systems that lack \e support in printf(1). * Issue #19: Removed stale references to atf-config and atf-run.
2014-10-30 17:14:07 +00:00
vars["srcdir"] = handle_srcdir(argv0, srcdir_arg).str();
Import atf-0.21: Released on October 23rd, 2014. * Restored the atf(7) manual page to serve as a reference to all the other manual pages shipped by ATF. * Added the -s flag to atf-sh to support specifying the shell interpreter to be used. * Removed ATF_WORKDIR. The only remaining consumers have been converted to use the standard TMPDIR environment variable. As a benefit, and because Kyua forces the TMPDIR to live within the test case's work directory, any stale files left behind by ATF will be automatically cleaned up. * Documented the environment variables recognized by each component in the relevant manual pages. This information was lost with the atf-config(1) removal. * Added a new "require.diskspace" metadata property to test cases so that they can specify the minimum amount of disk space required for the test to run. * Renamed the atf-{c,c++,sh}-api(3) manual pages to atf-{c,c++,sh}(3) for discoverability purposes. Symbolic links are provided for the time being to still make the old names visible. * Issue #5: Recommend the (expected, actual) idiom for calls to the test macros in the manual pages. * Issue #7: Stopped catching unhandled exceptions in atf-c++ tests. This propagates the crash to the caller, which in turn allows it to obtain proper debugging information. In particular, Kyua should now be able to extract a stacktrace pinpointing the problem. * Issue #8: Fixed atf-c/macros_test:use test failures spotted by the clang that ships with FreeBSD 11.0-CURRENT. * Issue #12: Improved documentation of atf-sh(3) and atf-check(1) by better explaining how they relate to each other. * Issue #14: Stopped setting 'set -e' in atf-sh. This setting was initially added as a way to enable a "strict" mode in the library and to make test cases fail fast when they run unprotected commands. However, doing so in the library is surprising as the responsibility of enabling 'set -e' should be on the user's code. Also, 'set -e' introduces inconsistent behavior on subshells and users do not expect that. * Issue #15: Fixed atf_utils_{fork,wait} to support nested calls. * Issue #16: Fixed test failures (by removing a long-standing hack) on systems that lack \e support in printf(1). * Issue #19: Removed stale references to atf-config and atf-run.
2014-10-30 17:14:07 +00:00
int errcode;
Import atf-0.21: Released on October 23rd, 2014. * Restored the atf(7) manual page to serve as a reference to all the other manual pages shipped by ATF. * Added the -s flag to atf-sh to support specifying the shell interpreter to be used. * Removed ATF_WORKDIR. The only remaining consumers have been converted to use the standard TMPDIR environment variable. As a benefit, and because Kyua forces the TMPDIR to live within the test case's work directory, any stale files left behind by ATF will be automatically cleaned up. * Documented the environment variables recognized by each component in the relevant manual pages. This information was lost with the atf-config(1) removal. * Added a new "require.diskspace" metadata property to test cases so that they can specify the minimum amount of disk space required for the test to run. * Renamed the atf-{c,c++,sh}-api(3) manual pages to atf-{c,c++,sh}(3) for discoverability purposes. Symbolic links are provided for the time being to still make the old names visible. * Issue #5: Recommend the (expected, actual) idiom for calls to the test macros in the manual pages. * Issue #7: Stopped catching unhandled exceptions in atf-c++ tests. This propagates the crash to the caller, which in turn allows it to obtain proper debugging information. In particular, Kyua should now be able to extract a stacktrace pinpointing the problem. * Issue #8: Fixed atf-c/macros_test:use test failures spotted by the clang that ships with FreeBSD 11.0-CURRENT. * Issue #12: Improved documentation of atf-sh(3) and atf-check(1) by better explaining how they relate to each other. * Issue #14: Stopped setting 'set -e' in atf-sh. This setting was initially added as a way to enable a "strict" mode in the library and to make test cases fail fast when they run unprotected commands. However, doing so in the library is surprising as the responsibility of enabling 'set -e' should be on the user's code. Also, 'set -e' introduces inconsistent behavior on subshells and users do not expect that. * Issue #15: Fixed atf_utils_{fork,wait} to support nested calls. * Issue #16: Fixed test failures (by removing a long-standing hack) on systems that lack \e support in printf(1). * Issue #19: Removed stale references to atf-config and atf-run.
2014-10-30 17:14:07 +00:00
tc_vector tcs;
if (lflag) {
if (argc > 0)
throw usage_error("Cannot provide test case names with -l");
Import atf-0.21: Released on October 23rd, 2014. * Restored the atf(7) manual page to serve as a reference to all the other manual pages shipped by ATF. * Added the -s flag to atf-sh to support specifying the shell interpreter to be used. * Removed ATF_WORKDIR. The only remaining consumers have been converted to use the standard TMPDIR environment variable. As a benefit, and because Kyua forces the TMPDIR to live within the test case's work directory, any stale files left behind by ATF will be automatically cleaned up. * Documented the environment variables recognized by each component in the relevant manual pages. This information was lost with the atf-config(1) removal. * Added a new "require.diskspace" metadata property to test cases so that they can specify the minimum amount of disk space required for the test to run. * Renamed the atf-{c,c++,sh}-api(3) manual pages to atf-{c,c++,sh}(3) for discoverability purposes. Symbolic links are provided for the time being to still make the old names visible. * Issue #5: Recommend the (expected, actual) idiom for calls to the test macros in the manual pages. * Issue #7: Stopped catching unhandled exceptions in atf-c++ tests. This propagates the crash to the caller, which in turn allows it to obtain proper debugging information. In particular, Kyua should now be able to extract a stacktrace pinpointing the problem. * Issue #8: Fixed atf-c/macros_test:use test failures spotted by the clang that ships with FreeBSD 11.0-CURRENT. * Issue #12: Improved documentation of atf-sh(3) and atf-check(1) by better explaining how they relate to each other. * Issue #14: Stopped setting 'set -e' in atf-sh. This setting was initially added as a way to enable a "strict" mode in the library and to make test cases fail fast when they run unprotected commands. However, doing so in the library is surprising as the responsibility of enabling 'set -e' should be on the user's code. Also, 'set -e' introduces inconsistent behavior on subshells and users do not expect that. * Issue #15: Fixed atf_utils_{fork,wait} to support nested calls. * Issue #16: Fixed test failures (by removing a long-standing hack) on systems that lack \e support in printf(1). * Issue #19: Removed stale references to atf-config and atf-run.
2014-10-30 17:14:07 +00:00
init_tcs(add_tcs, tcs, vars);
errcode = list_tcs(tcs);
} else {
Import atf-0.21: Released on October 23rd, 2014. * Restored the atf(7) manual page to serve as a reference to all the other manual pages shipped by ATF. * Added the -s flag to atf-sh to support specifying the shell interpreter to be used. * Removed ATF_WORKDIR. The only remaining consumers have been converted to use the standard TMPDIR environment variable. As a benefit, and because Kyua forces the TMPDIR to live within the test case's work directory, any stale files left behind by ATF will be automatically cleaned up. * Documented the environment variables recognized by each component in the relevant manual pages. This information was lost with the atf-config(1) removal. * Added a new "require.diskspace" metadata property to test cases so that they can specify the minimum amount of disk space required for the test to run. * Renamed the atf-{c,c++,sh}-api(3) manual pages to atf-{c,c++,sh}(3) for discoverability purposes. Symbolic links are provided for the time being to still make the old names visible. * Issue #5: Recommend the (expected, actual) idiom for calls to the test macros in the manual pages. * Issue #7: Stopped catching unhandled exceptions in atf-c++ tests. This propagates the crash to the caller, which in turn allows it to obtain proper debugging information. In particular, Kyua should now be able to extract a stacktrace pinpointing the problem. * Issue #8: Fixed atf-c/macros_test:use test failures spotted by the clang that ships with FreeBSD 11.0-CURRENT. * Issue #12: Improved documentation of atf-sh(3) and atf-check(1) by better explaining how they relate to each other. * Issue #14: Stopped setting 'set -e' in atf-sh. This setting was initially added as a way to enable a "strict" mode in the library and to make test cases fail fast when they run unprotected commands. However, doing so in the library is surprising as the responsibility of enabling 'set -e' should be on the user's code. Also, 'set -e' introduces inconsistent behavior on subshells and users do not expect that. * Issue #15: Fixed atf_utils_{fork,wait} to support nested calls. * Issue #16: Fixed test failures (by removing a long-standing hack) on systems that lack \e support in printf(1). * Issue #19: Removed stale references to atf-config and atf-run.
2014-10-30 17:14:07 +00:00
if (argc == 0)
throw usage_error("Must provide a test case name");
Import atf-0.21: Released on October 23rd, 2014. * Restored the atf(7) manual page to serve as a reference to all the other manual pages shipped by ATF. * Added the -s flag to atf-sh to support specifying the shell interpreter to be used. * Removed ATF_WORKDIR. The only remaining consumers have been converted to use the standard TMPDIR environment variable. As a benefit, and because Kyua forces the TMPDIR to live within the test case's work directory, any stale files left behind by ATF will be automatically cleaned up. * Documented the environment variables recognized by each component in the relevant manual pages. This information was lost with the atf-config(1) removal. * Added a new "require.diskspace" metadata property to test cases so that they can specify the minimum amount of disk space required for the test to run. * Renamed the atf-{c,c++,sh}-api(3) manual pages to atf-{c,c++,sh}(3) for discoverability purposes. Symbolic links are provided for the time being to still make the old names visible. * Issue #5: Recommend the (expected, actual) idiom for calls to the test macros in the manual pages. * Issue #7: Stopped catching unhandled exceptions in atf-c++ tests. This propagates the crash to the caller, which in turn allows it to obtain proper debugging information. In particular, Kyua should now be able to extract a stacktrace pinpointing the problem. * Issue #8: Fixed atf-c/macros_test:use test failures spotted by the clang that ships with FreeBSD 11.0-CURRENT. * Issue #12: Improved documentation of atf-sh(3) and atf-check(1) by better explaining how they relate to each other. * Issue #14: Stopped setting 'set -e' in atf-sh. This setting was initially added as a way to enable a "strict" mode in the library and to make test cases fail fast when they run unprotected commands. However, doing so in the library is surprising as the responsibility of enabling 'set -e' should be on the user's code. Also, 'set -e' introduces inconsistent behavior on subshells and users do not expect that. * Issue #15: Fixed atf_utils_{fork,wait} to support nested calls. * Issue #16: Fixed test failures (by removing a long-standing hack) on systems that lack \e support in printf(1). * Issue #19: Removed stale references to atf-config and atf-run.
2014-10-30 17:14:07 +00:00
else if (argc > 1)
throw usage_error("Cannot provide more than one test case name");
Import atf-0.21: Released on October 23rd, 2014. * Restored the atf(7) manual page to serve as a reference to all the other manual pages shipped by ATF. * Added the -s flag to atf-sh to support specifying the shell interpreter to be used. * Removed ATF_WORKDIR. The only remaining consumers have been converted to use the standard TMPDIR environment variable. As a benefit, and because Kyua forces the TMPDIR to live within the test case's work directory, any stale files left behind by ATF will be automatically cleaned up. * Documented the environment variables recognized by each component in the relevant manual pages. This information was lost with the atf-config(1) removal. * Added a new "require.diskspace" metadata property to test cases so that they can specify the minimum amount of disk space required for the test to run. * Renamed the atf-{c,c++,sh}-api(3) manual pages to atf-{c,c++,sh}(3) for discoverability purposes. Symbolic links are provided for the time being to still make the old names visible. * Issue #5: Recommend the (expected, actual) idiom for calls to the test macros in the manual pages. * Issue #7: Stopped catching unhandled exceptions in atf-c++ tests. This propagates the crash to the caller, which in turn allows it to obtain proper debugging information. In particular, Kyua should now be able to extract a stacktrace pinpointing the problem. * Issue #8: Fixed atf-c/macros_test:use test failures spotted by the clang that ships with FreeBSD 11.0-CURRENT. * Issue #12: Improved documentation of atf-sh(3) and atf-check(1) by better explaining how they relate to each other. * Issue #14: Stopped setting 'set -e' in atf-sh. This setting was initially added as a way to enable a "strict" mode in the library and to make test cases fail fast when they run unprotected commands. However, doing so in the library is surprising as the responsibility of enabling 'set -e' should be on the user's code. Also, 'set -e' introduces inconsistent behavior on subshells and users do not expect that. * Issue #15: Fixed atf_utils_{fork,wait} to support nested calls. * Issue #16: Fixed test failures (by removing a long-standing hack) on systems that lack \e support in printf(1). * Issue #19: Removed stale references to atf-config and atf-run.
2014-10-30 17:14:07 +00:00
INV(argc == 1);
Import atf-0.21: Released on October 23rd, 2014. * Restored the atf(7) manual page to serve as a reference to all the other manual pages shipped by ATF. * Added the -s flag to atf-sh to support specifying the shell interpreter to be used. * Removed ATF_WORKDIR. The only remaining consumers have been converted to use the standard TMPDIR environment variable. As a benefit, and because Kyua forces the TMPDIR to live within the test case's work directory, any stale files left behind by ATF will be automatically cleaned up. * Documented the environment variables recognized by each component in the relevant manual pages. This information was lost with the atf-config(1) removal. * Added a new "require.diskspace" metadata property to test cases so that they can specify the minimum amount of disk space required for the test to run. * Renamed the atf-{c,c++,sh}-api(3) manual pages to atf-{c,c++,sh}(3) for discoverability purposes. Symbolic links are provided for the time being to still make the old names visible. * Issue #5: Recommend the (expected, actual) idiom for calls to the test macros in the manual pages. * Issue #7: Stopped catching unhandled exceptions in atf-c++ tests. This propagates the crash to the caller, which in turn allows it to obtain proper debugging information. In particular, Kyua should now be able to extract a stacktrace pinpointing the problem. * Issue #8: Fixed atf-c/macros_test:use test failures spotted by the clang that ships with FreeBSD 11.0-CURRENT. * Issue #12: Improved documentation of atf-sh(3) and atf-check(1) by better explaining how they relate to each other. * Issue #14: Stopped setting 'set -e' in atf-sh. This setting was initially added as a way to enable a "strict" mode in the library and to make test cases fail fast when they run unprotected commands. However, doing so in the library is surprising as the responsibility of enabling 'set -e' should be on the user's code. Also, 'set -e' introduces inconsistent behavior on subshells and users do not expect that. * Issue #15: Fixed atf_utils_{fork,wait} to support nested calls. * Issue #16: Fixed test failures (by removing a long-standing hack) on systems that lack \e support in printf(1). * Issue #19: Removed stale references to atf-config and atf-run.
2014-10-30 17:14:07 +00:00
init_tcs(add_tcs, tcs, vars);
errcode = run_tc(tcs, argv[0], resfile);
}
for (tc_vector::iterator iter = tcs.begin(); iter != tcs.end(); iter++) {
impl::tc* tc = *iter;
delete tc;
}
return errcode;
}
Import atf-0.21: Released on October 23rd, 2014. * Restored the atf(7) manual page to serve as a reference to all the other manual pages shipped by ATF. * Added the -s flag to atf-sh to support specifying the shell interpreter to be used. * Removed ATF_WORKDIR. The only remaining consumers have been converted to use the standard TMPDIR environment variable. As a benefit, and because Kyua forces the TMPDIR to live within the test case's work directory, any stale files left behind by ATF will be automatically cleaned up. * Documented the environment variables recognized by each component in the relevant manual pages. This information was lost with the atf-config(1) removal. * Added a new "require.diskspace" metadata property to test cases so that they can specify the minimum amount of disk space required for the test to run. * Renamed the atf-{c,c++,sh}-api(3) manual pages to atf-{c,c++,sh}(3) for discoverability purposes. Symbolic links are provided for the time being to still make the old names visible. * Issue #5: Recommend the (expected, actual) idiom for calls to the test macros in the manual pages. * Issue #7: Stopped catching unhandled exceptions in atf-c++ tests. This propagates the crash to the caller, which in turn allows it to obtain proper debugging information. In particular, Kyua should now be able to extract a stacktrace pinpointing the problem. * Issue #8: Fixed atf-c/macros_test:use test failures spotted by the clang that ships with FreeBSD 11.0-CURRENT. * Issue #12: Improved documentation of atf-sh(3) and atf-check(1) by better explaining how they relate to each other. * Issue #14: Stopped setting 'set -e' in atf-sh. This setting was initially added as a way to enable a "strict" mode in the library and to make test cases fail fast when they run unprotected commands. However, doing so in the library is surprising as the responsibility of enabling 'set -e' should be on the user's code. Also, 'set -e' introduces inconsistent behavior on subshells and users do not expect that. * Issue #15: Fixed atf_utils_{fork,wait} to support nested calls. * Issue #16: Fixed test failures (by removing a long-standing hack) on systems that lack \e support in printf(1). * Issue #19: Removed stale references to atf-config and atf-run.
2014-10-30 17:14:07 +00:00
} // anonymous namespace
namespace atf {
namespace tests {
Import atf-0.21: Released on October 23rd, 2014. * Restored the atf(7) manual page to serve as a reference to all the other manual pages shipped by ATF. * Added the -s flag to atf-sh to support specifying the shell interpreter to be used. * Removed ATF_WORKDIR. The only remaining consumers have been converted to use the standard TMPDIR environment variable. As a benefit, and because Kyua forces the TMPDIR to live within the test case's work directory, any stale files left behind by ATF will be automatically cleaned up. * Documented the environment variables recognized by each component in the relevant manual pages. This information was lost with the atf-config(1) removal. * Added a new "require.diskspace" metadata property to test cases so that they can specify the minimum amount of disk space required for the test to run. * Renamed the atf-{c,c++,sh}-api(3) manual pages to atf-{c,c++,sh}(3) for discoverability purposes. Symbolic links are provided for the time being to still make the old names visible. * Issue #5: Recommend the (expected, actual) idiom for calls to the test macros in the manual pages. * Issue #7: Stopped catching unhandled exceptions in atf-c++ tests. This propagates the crash to the caller, which in turn allows it to obtain proper debugging information. In particular, Kyua should now be able to extract a stacktrace pinpointing the problem. * Issue #8: Fixed atf-c/macros_test:use test failures spotted by the clang that ships with FreeBSD 11.0-CURRENT. * Issue #12: Improved documentation of atf-sh(3) and atf-check(1) by better explaining how they relate to each other. * Issue #14: Stopped setting 'set -e' in atf-sh. This setting was initially added as a way to enable a "strict" mode in the library and to make test cases fail fast when they run unprotected commands. However, doing so in the library is surprising as the responsibility of enabling 'set -e' should be on the user's code. Also, 'set -e' introduces inconsistent behavior on subshells and users do not expect that. * Issue #15: Fixed atf_utils_{fork,wait} to support nested calls. * Issue #16: Fixed test failures (by removing a long-standing hack) on systems that lack \e support in printf(1). * Issue #19: Removed stale references to atf-config and atf-run.
2014-10-30 17:14:07 +00:00
int run_tp(int, char**, void (*)(tc_vector&));
}
}
int
Import atf-0.21: Released on October 23rd, 2014. * Restored the atf(7) manual page to serve as a reference to all the other manual pages shipped by ATF. * Added the -s flag to atf-sh to support specifying the shell interpreter to be used. * Removed ATF_WORKDIR. The only remaining consumers have been converted to use the standard TMPDIR environment variable. As a benefit, and because Kyua forces the TMPDIR to live within the test case's work directory, any stale files left behind by ATF will be automatically cleaned up. * Documented the environment variables recognized by each component in the relevant manual pages. This information was lost with the atf-config(1) removal. * Added a new "require.diskspace" metadata property to test cases so that they can specify the minimum amount of disk space required for the test to run. * Renamed the atf-{c,c++,sh}-api(3) manual pages to atf-{c,c++,sh}(3) for discoverability purposes. Symbolic links are provided for the time being to still make the old names visible. * Issue #5: Recommend the (expected, actual) idiom for calls to the test macros in the manual pages. * Issue #7: Stopped catching unhandled exceptions in atf-c++ tests. This propagates the crash to the caller, which in turn allows it to obtain proper debugging information. In particular, Kyua should now be able to extract a stacktrace pinpointing the problem. * Issue #8: Fixed atf-c/macros_test:use test failures spotted by the clang that ships with FreeBSD 11.0-CURRENT. * Issue #12: Improved documentation of atf-sh(3) and atf-check(1) by better explaining how they relate to each other. * Issue #14: Stopped setting 'set -e' in atf-sh. This setting was initially added as a way to enable a "strict" mode in the library and to make test cases fail fast when they run unprotected commands. However, doing so in the library is surprising as the responsibility of enabling 'set -e' should be on the user's code. Also, 'set -e' introduces inconsistent behavior on subshells and users do not expect that. * Issue #15: Fixed atf_utils_{fork,wait} to support nested calls. * Issue #16: Fixed test failures (by removing a long-standing hack) on systems that lack \e support in printf(1). * Issue #19: Removed stale references to atf-config and atf-run.
2014-10-30 17:14:07 +00:00
impl::run_tp(int argc, char** argv, void (*add_tcs)(tc_vector&))
{
Import atf-0.21: Released on October 23rd, 2014. * Restored the atf(7) manual page to serve as a reference to all the other manual pages shipped by ATF. * Added the -s flag to atf-sh to support specifying the shell interpreter to be used. * Removed ATF_WORKDIR. The only remaining consumers have been converted to use the standard TMPDIR environment variable. As a benefit, and because Kyua forces the TMPDIR to live within the test case's work directory, any stale files left behind by ATF will be automatically cleaned up. * Documented the environment variables recognized by each component in the relevant manual pages. This information was lost with the atf-config(1) removal. * Added a new "require.diskspace" metadata property to test cases so that they can specify the minimum amount of disk space required for the test to run. * Renamed the atf-{c,c++,sh}-api(3) manual pages to atf-{c,c++,sh}(3) for discoverability purposes. Symbolic links are provided for the time being to still make the old names visible. * Issue #5: Recommend the (expected, actual) idiom for calls to the test macros in the manual pages. * Issue #7: Stopped catching unhandled exceptions in atf-c++ tests. This propagates the crash to the caller, which in turn allows it to obtain proper debugging information. In particular, Kyua should now be able to extract a stacktrace pinpointing the problem. * Issue #8: Fixed atf-c/macros_test:use test failures spotted by the clang that ships with FreeBSD 11.0-CURRENT. * Issue #12: Improved documentation of atf-sh(3) and atf-check(1) by better explaining how they relate to each other. * Issue #14: Stopped setting 'set -e' in atf-sh. This setting was initially added as a way to enable a "strict" mode in the library and to make test cases fail fast when they run unprotected commands. However, doing so in the library is surprising as the responsibility of enabling 'set -e' should be on the user's code. Also, 'set -e' introduces inconsistent behavior on subshells and users do not expect that. * Issue #15: Fixed atf_utils_{fork,wait} to support nested calls. * Issue #16: Fixed test failures (by removing a long-standing hack) on systems that lack \e support in printf(1). * Issue #19: Removed stale references to atf-config and atf-run.
2014-10-30 17:14:07 +00:00
try {
set_program_name(argv[0]);
return ::safe_main(argc, argv, add_tcs);
} catch (const usage_error& e) {
std::cerr
<< Program_Name << ": ERROR: " << e.what() << '\n'
<< Program_Name << ": See atf-test-program(1) for usage details.\n";
return EXIT_FAILURE;
}
}