The severity now depends only on the setting of SUBST_NOOP_OK. Right now
this means that some former warnings will be reported as info only, but
that will change after switching the default of SUBST_NOOP_OK after
2020Q1. Then they will all be reported as warnings, followed by the final
error saying that the pattern has no effect.
This change makes it easier to detect inconsistencies and outdated
definitions, for example by setting the global SUBST_NOOP_OK=no and
redefining WARNING_MSG to actuall fail.
The diff program is only used to produce informative output in the build
logs, nevertheless its output might be translated if there are lines that
do not end with a newline.
The default value of SUBST_MESSAGE is based on SUBST_FILES, and that
variable may use the :sh modifier to list files from WRKSRC, which may
not exist at load time.
In the case of pkglocaledir, the SUBST_FILES are generated by a shell
command. That command assumes that the WRKDIR already exists. Therefore
SUBST_FILES must be evaluated as late as possible.
See mk/configure/replace-localedir.mk; an example package that fails is
devel/gettext-tools.
In a bulk build with very strict settings (WARNING_MSG fails, as well as
no-op substitutions), it became clear that nearly all of the cases where
SUBST didn't replace anything were bugs in the package definition.
Most of them were just outdated, which is no surprise given that some
packages are already over 20 years old.
For backwards compatibility, SUBST_NOOP_OK defaults to "yes" right now.
After correcting the affected packages, the default will change to "no".
All variables that are used or defined in the file are now listed in the
_VARGROUPS section.
The "is text file" command variable has been renamed since pkglint
thought the former variable name would specify a filename, not a shell
command.
The "is text file" command has been rewritten to only rely on tr(1)
instead of both tr(1) and wc(1). This makes it both simpler and maybe
also a little faster, since the file only has to be read once.
The SUBST_TARGETS variable has been removed since it is used nowhere
else. To get the list of all subst targets (should that ever be
necessary), use the expression ${SUBST_CLASSES:S,^,subst-,}.
Up to now, there was a central list of variable name patterns that
defined whether a variable was printed as a sorted list, as a list or as
a single value.
Now each variable group decides on its own which of the variables are
printed in which way, using the usual glob patterns. This is more
flexible since different files sometimes differ in their naming
conventions.
Two variable groups are added: license (for everything related to
LICENSE) and go (for lang/go).
Up to now, using subst.mk may have led to file corruption during active
package development. This happened when a sed(1) command had a syntax
error, in which case the whole sed(1) command was terminated, leaving an
empty original file behind.
This commit changes that behavior by applying the sed(1) commands to
the original file and saving the result in a temporary file. Only
after that succeeded is the original file overwritten.
During this rewrite, SUBST_POSTCMD has been removed, since it was
only used in one place (mk/wrapper), and since it relied on the exact
sequence of the internal commands. No package in either main pkgsrc
or pkgsrc-wip uses this variable right now.
This way, file modifications cannot go unnoticed and at least appear in
the build logs. While here, fixed an ambiguity in the description of
SUBST_MESSAGE.
has proven too unreliable in the past. For example, some Makefile.in
files were classified as "Quake I or II world or extension", just
because they happen to start with the letters "PACK". This method was
also subject to subtle differences in the locale.
The new method counts the number of NUL bytes in the file. It does not
depend on the locale settings. The -c option of wc(1) counts bytes, not
characters, and tr(1), which may interpret multibyte sequences, is
protected by LC_ALL. It should also work with the historical
implementations of tr(1) that could not handle NUL bytes and discarded
them, since this is exactly the intention.
See also:
* http://mail-index.netbsd.org/tech-pkg/2006/07/05/0000.html
* PR 37793
is a new target "show-all" that fits to the existing "debug",
"show-tools", "show-vars" targets. It prints a list of the variables
that make up the public interface to pkgsrc. Running this target is
especially useful if you want to do some things, you know that they must
have been implemented but you don't know what it is called. It also
shows the "class" of a variable (user-defined, package-defined,
system-defined).
"gdk-pixbuf/Makefile.in" from the "gtk2" package as "libtool library file".
Change the textfile check to accept any file which file(1) identifies
as "<whatever>libtool<whateverelse>" to fix the build of the
"gtk2" package (and probably other packages).
PKG_FAIL_REASON in that case. It didn't have an effect anyway for normal
builds, since subst.mk is included after checking PKG_FAIL_REASON.
Discussed with jlam.
evaluated. Now the SUBST_MESSAGE is only printed once when the
substitution is actually done. Before this change it had been printed
also when the subst-<class> target had been invoked a second time, but
the substitution didn't take place again, which had confused me. Also,
converted the code to use ${WARNING_MSG} and ${STEP_MSG}.
SUBST_MESSAGE is printed. The ones that have been defined with "quotes"
in their Makefiles are printed with quotes (of course). This is the
consequence of the design pattern "quote-exactly-where-necessary", which
in fact should be have been applied to pkgsrc as a whole, but still isn't.
detection whether a given file was a text file or some other file had
been unreliable. In the recent bulk builds, all of the warnings that had
appeared because of that unreliable detection had been false positives.
variables that use the :sh modifier. This still causes expansion to only
happen when referenced, and has the advantage of being :Q-safe.
Bring back the changes from revision 1.19 of mk/subst.mk now that the
problem noted above has been fixed. This passes the buildlink-unwrap
regression test.
