Commit Graph

29 Commits

Author SHA1 Message Date
John Polstra
fc29ad10fd Fixed pkg_create to eliminate the limits on the size of the PLIST.
Formerly, there were limits on both the number of files (hard-coded into
the program) and the number of characters (because of the ARG_MAX limit
in exec(2)).  In this new version, the filenames are passed to tar through
a pipe, using tar's "-T" option, rather than on the command line.
1996-02-06 22:49:11 +00:00
Jordan K. Hubbard
36b7319742 Bring forward my changes from 2.1 1995-10-25 15:38:37 +00:00
Jordan K. Hubbard
82f99dbbda Bump the number of allowable args by a factor of 4. I just scanned
all possible packing lists and found that at least one of them is
pretty large!
1995-05-10 22:33:55 +00:00
Jordan K. Hubbard
c63a4dc13b Re-work make_dist() routine to build the tar command more reasonably.
This should eliminate the problem with truncated packages.
1995-05-10 20:46:06 +00:00
Jordan K. Hubbard
18a7c7704f Add a great deal more error checking to various things. 1995-04-26 15:06:58 +00:00
Jordan K. Hubbard
80206f052f Yikes! How did that typo slip past! I'd swear I compiled it. Oh well,
thanks, Satoshi!

Embarrassed by:	asami
1995-04-26 12:37:46 +00:00
Jordan K. Hubbard
d14afa6dd2 As per Bruce's advice, use sysconf to get the max argument size and
dynamically allocate that much space, also using snprintf() and strncat()
to do proper bounds checking.
1995-04-26 11:43:20 +00:00
Jordan K. Hubbard
1c491c8eb8 Use ARG_MAX instead of _POSIX_ARG_MAX: 4K is too small!
This should stop the SIGBUS errors in pkg_create for some packages.
1995-04-24 21:50:11 +00:00
Jordan K. Hubbard
2a20a9edde Stage almost nothing through the playpen now - this should speed pkg_creates
up by a couple of factors, as well as decrease the amount of temporary space
required to build a package down to almost nothing.
1995-04-22 14:55:07 +00:00
Jordan K. Hubbard
cbf098db0e Second round:
1. Make paths work correctly.
2. Make pkg_add generally more robust in the face of failure.
3. Make the depend messages come out on stderr or stdout, but not both
   interspersed! :-)
1995-04-22 13:58:24 +00:00
Jordan K. Hubbard
adbcdd9cce Argh - missed a printf! That should be the last of plist-spewers. 1995-04-22 01:20:13 +00:00
Jordan K. Hubbard
922a707e02 Don't spew garbage into the packing list when PlistOnly and verbose are
both on.  Whoops!
1995-04-22 00:59:33 +00:00
Jordan K. Hubbard
3a5e372a3d 1. Add proper dependencies to the library in my Makefiles. This was pointed
out by Bruce.
2. Add a "feature" to pkg_create (OK, OK, it's a miserable hack!) to get
   it to dump its internal packing list out so that the `fake-pkg' rule in
   bsd.port.mk can generate a more meaningful packing list.
1995-04-22 00:03:18 +00:00
Jordan K. Hubbard
1b164cce2b Neither of these tools really told you what you needed to know neatly
and succinctly in verbose mode.  I hope people find this more to their
liking.
1995-04-10 08:01:52 +00:00
Jordan K. Hubbard
6c5325bf0e Add true support for dependency lists.
1. pkg_create now has a -P argument for specifying dependencies on the
   command line.
2. pkg_add will honor dependencies and chain-load them automatically if
   it finds the required package(s) in the same directory as the package
   that is being loaded.  For best results, install packages from a directory
   containing all the packages you'll possibly need
   (like /usr/ports/packages/all).

2 remaining flaws:

1. pkg_add looks in one place (where you were when you loaded the primary
   pkg) for depended packages.  If you can come up with a search path scheme
   that's not a total hack - be my guest!

2. Recursive dependency expansion can result in the name of a dep being
   listed more than once.  This doesn't bother pkg_add since it checks
   for package existance with pkg_info and will skip already-loaded packages.
   I don't know how/if pkg_delete handles this yet, however.  I need to look
   into it.
1995-04-09 15:05:01 +00:00
Jordan K. Hubbard
8f396ebfb6 Many of John T. Kohl's patches from NetBSD. Thanks, John!
Submitted by:	jkohl
1994-12-06 00:51:50 +00:00
Jordan K. Hubbard
07b5b56eba Don't create packages with bogus @name lines. 1994-11-17 10:54:11 +00:00
Jordan K. Hubbard
69bb95d933 This should do it - I forgot about pkg_info. 1994-10-14 05:55:57 +00:00
Jordan K. Hubbard
92fdc4d3b4 My last commit to bsd.port.mk made me think - `pkg_create -p ...' should
override the PLIST file's own first @cwd.  Make this so.
1994-10-04 15:54:39 +00:00
Adam David
d70574cd96 Allow creation of absolute package names
typo index() --> rindex()  in order to split correct suffix
Fix @srcdir so it actually works
Submitted by:	adam
1994-08-29 16:31:38 +00:00
Jordan K. Hubbard
b6532a528a Change in_plist() call to find_plist(). 1994-05-25 06:24:41 +00:00
Andrew Moore
6a87350c4e Add `-h' option to force tar to follow symbolic links 1994-05-19 18:27:41 +00:00
Jordan K. Hubbard
d4473b1c59 Doc new % sequence expansion for command execution.
Fixed bad verbose message.
1993-09-18 03:39:21 +00:00
Jordan K. Hubbard
77277aa67e Whoops - got the exclude flag wrong, try again. 1993-09-12 20:56:40 +00:00
Jordan K. Hubbard
7cb39e249b Added and doc'd -X flag to pkg_create that allows you to specify an exclude
list for the final package.
1993-09-12 20:45:31 +00:00
Jordan K. Hubbard
e0ab6cb9c7 Made staging area configurable rather than assuming /tmp (-t) 1993-09-05 04:54:24 +00:00
Jordan K. Hubbard
14cb3fbecd Fixed bug in pkg_create that affected Rich's attempts to package emacs 1993-09-04 18:19:41 +00:00
Jordan K. Hubbard
a9c42a2388 Lots of misc tweaks, support for arbitrary separators in pkg_info, more
intelligent name handling in pkg_create.  Most of these files are changed
because of rcsid's being different in my cvs tree and freefall's (foo).
1993-09-03 23:01:17 +00:00
Jordan K. Hubbard
6d946b2e52 The release version of my package install suite. Please see man pages
for info.
1993-08-26 01:19:55 +00:00