06b9b3e0ad
Quite a lot of churn on style, but lots of good work refactoring complicated functions and lots more unit-tests. Thanks mostly to rillig at NetBSD Some interesting entries from ChangeLog o .MAKE.{UID,GID} represent uid and gid running make. o allow env var MAKE_OBJDIR_CHECK_WRITABLE=no to skip writable checks in InitObjdir. Explicit .OBJDIR target always allows read-only directory. o add more unit tests for META MODE Merge commit '8e11a9b4250be3c3379c45fa820bff78d99d5946' into main Change-Id: I464fd4c013067f0915671c1ccc96d2d8090b2b9c
78 lines
2.8 KiB
Makefile
78 lines
2.8 KiB
Makefile
# $NetBSD: var-op-default.mk,v 1.3 2020/12/07 21:35:43 rillig Exp $
|
|
#
|
|
# Tests for the ?= variable assignment operator, which only assigns
|
|
# if the variable is still undefined.
|
|
|
|
# The variable VAR is not defined yet. Therefore it gets the default value
|
|
# from the variable assignment.
|
|
VAR?= default value
|
|
.if ${VAR} != "default value"
|
|
. error
|
|
.endif
|
|
|
|
# At this point, the variable 'VAR' is already defined. The '?=' therefore
|
|
# ignores the new variable value, preserving the previous "default value".
|
|
VAR?= ignored
|
|
.if ${VAR} != "default value"
|
|
. error
|
|
.endif
|
|
|
|
# The '?=' operator only checks whether the variable is defined or not.
|
|
# An empty variable is defined, therefore the '?=' operator does nothing.
|
|
EMPTY= # empty
|
|
EMPTY?= ignored
|
|
.if ${EMPTY} != ""
|
|
. error
|
|
.endif
|
|
|
|
# The .for loop is described in the manual page as if it would operate on
|
|
# variables. This is not entirely true. Instead, each occurrence of an
|
|
# expression $i or ${i} or ${i:...} is substituted with ${:Uloop-value}.
|
|
# This comes very close to the description, the only difference is that
|
|
# there is never an actual variable named 'i' involved.
|
|
#
|
|
# Because there is not really a variable named 'i', the '?=' operator
|
|
# performs the variable assignment, resulting in $i == "default".
|
|
.for i in loop-value
|
|
i?= default
|
|
.endfor
|
|
.if ${i} != "default"
|
|
. error
|
|
.endif
|
|
|
|
# At the point where the '?=' operator checks whether the variable exists,
|
|
# it expands the variable name exactly once. Therefore both 'VAR.param'
|
|
# and 'VAR.${param}' expand to 'VAR.param', and the second '?=' assignment
|
|
# has no effect.
|
|
#
|
|
# Since 2000.05.11.07.43.42 it has been possible to use nested variable
|
|
# expressions in variable names, which made make much more versatile.
|
|
# On 2008.03.31.00.12.21, this particular case of the '?=' operator has been
|
|
# fixed. Before, the '?=' operator had not expanded the variable name
|
|
# 'VAR.${:Uparam}' to see whether the variable already existed. Since that
|
|
# variable didn't exist (and variables with '$' in their name are particularly
|
|
# fragile), the variable assignment with "not used" was performed, and only
|
|
# during that, the variable name was expanded.
|
|
VAR.param= already defined
|
|
VAR.${:Uparam}?= not used
|
|
.if ${VAR.param} != "already defined"
|
|
. error
|
|
.endif
|
|
|
|
# Now demonstrate that the variable name is indeed expanded exactly once.
|
|
# This is tricky to measure correctly since there are many inconsistencies
|
|
# in and around the code that expands variable expressions in the various
|
|
# places where variable expressions can occur. If in doubt, enable the
|
|
# following debug flags to see what happens:
|
|
#.MAKEFLAGS: -dcpv
|
|
EXPAND_NAME= EXPAND.$$$$ # The full variable name is EXPAND.$$
|
|
PARAM= $$$$
|
|
EXPAND.${PARAM}?= value with param
|
|
.if ${${EXPAND_NAME}} != "value with param"
|
|
. error
|
|
.endif
|
|
.MAKEFLAGS: -d0
|
|
|
|
all:
|
|
@:;
|