[PATCH] gnulib-tool.py: Follow gnulib-tool changes, part 69.

2024-03-22 Thread Collin Funk
This patch should hopefully stop gnulib-tool.py from mangling .gitignore files like it did previously. I think I did a decent job translating the join calls into Python sets, but any criticism is welcome. I tried using gnulib-tool.py to bootstrap GNU Inetutils and it handles the .gitignore files p

Re: [PATCH] gnulib-tool.py: Don't remove newlines in testdir's Automake snippets.

2024-03-22 Thread Bruno Haible
Thanks! Both patches applied. > I find it easier to follow this: > > emit += 'AM_PROG_CC_C_O\n' > emit += '\n' > > rather than this: > > emit += 'AM_PROG_CC_C_O\n\n' I agree. Bruno

[PATCH] gnulib-tool.py: Don't remove newlines in testdir's Automake snippets.

2024-03-22 Thread Collin Funk
This patch should *hopefully* cover the last newline issue in test directories. It seems that gnulib-tool.py strips the whitespace from Automake snippets where gnulib-tool.sh doesn't. This can be seen by the following diff: $ diff -u ./test-create-megatestdir-1.result/ALL/configure.ac tmp500214-

[PATCH] gnulib-tool.py: Fix empty newline output in test directories.

2024-03-22 Thread Collin Funk
This patch just fixes two simple newline differences in test directories. $ diff -u ./test-create-megatestdir-2.result/fopen-gnu/configure.ac tmp364942-result/fopen-gnu/configure.ac --- ./test-create-megatestdir-2.result/fopen-gnu/configure.ac 2024-03-22 11:23:00.239344218 -0700 +++ tmp364942-

Re: gnulib-tool.py: Fix an error in --create-testdir

2024-03-22 Thread Collin Funk
Hi Bruno, On 3/22/24 8:02 AM, Bruno Haible wrote: > This test error > > $ GNULIB_TOOL_IMPL=py ./test-create-testdir-1.sh > configure.ac:8: installing 'build-aux/compile' > configure.ac:4: installing 'build-aux/install-sh' > configure.ac:4: installing 'build-aux/missing' > gllib/Makefile.am: inst

Re: [PATCH] gnulib-tool.py: Fix unconditional Automake snippets for non-tests.

2024-03-22 Thread Bruno Haible
Hi Collin, > filter_filelist will join an empty list. In > GLModule.getAutomakeSnippet_Unconditional we split by the same > separator which results in a list with an empty string. ... which is then (at least once) converted to a set with an empty string and back to a list with an empty string. Pr

Re: test suite problem

2024-03-22 Thread Bruno Haible
Hi Collin, > I assume you are using Autoconf 2.72? I don't see an easy way to fix > this and Autoconf is trivial to install so it is probably just worth > mentioning in the README. > > $ /usr/bin/autoconf --version > autoconf (GNU Autoconf) 2.71 > $ env GNULIB_TOOL_IMPL=py AUTOCONF=/usr/bin/autoc

gnulib-tool.sh: Produce same diagnostics regardless of Automake version

2024-03-22 Thread Bruno Haible
With Automake 1.16.3 in $PATH, "gnulib-tool --create-testdir ..." produces more diagnostics than with Automake 1.16.5 in $PATH. It is better if the diagnostics are the same: better for the user, but also for the test suite. This patch does it. 2024-03-22 Bruno Haible gnulib-tool.sh:

gnulib-tool: Remove test-driver.{orig,rej} if patching succeeded

2024-03-22 Thread Bruno Haible
When running "gnulib-tool --create-testdir ..." with Automake 1.16.3 in $PATH, two files build-aux/test-driver.orig and build-aux/test-driver.rej are created in the destination directory. This is not useful, because - the second 'patch' command succeeded, - the user is not interested in debuggi

gnulib-tool.py: Fix an error in --create-testdir

2024-03-22 Thread Bruno Haible
This test error $ GNULIB_TOOL_IMPL=py ./test-create-testdir-1.sh configure.ac:8: installing 'build-aux/compile' configure.ac:4: installing 'build-aux/install-sh' configure.ac:4: installing 'build-aux/missing' gllib/Makefile.am: installing 'build-aux/depcomp' parallel-tests: installing 'build-aux/

Re: [PATCH] gnulib-tool.py: Fix unconditional Automake snippets for non-tests.

2024-03-22 Thread Collin Funk
Hi Bruno, Here is a patch with some updates now that I looked into the issue a bit more. I've also added the comments and fixed the build-aux issue you noticed. On 3/21/24 3:20 PM, Collin Funk wrote: > This is > because filter_filelist can return a non-empty list with an empty > string: This isn