bug-texinfo
[Top][All Lists]
Advanced

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: Texinfo 7.1.0.90 pretest results [OpenBSD 7.5]


From: Patrice Dumas
Subject: Re: Texinfo 7.1.0.90 pretest results [OpenBSD 7.5]
Date: Mon, 10 Jun 2024 14:56:25 +0200

On Mon, Jun 10, 2024 at 01:27:58PM +0200, Bruno Haible wrote:
> Gavin Smith wrote:
> > I tried to fix one problem on the master branch (commit 2ae196b9807a,
> > 2024-06-09).
> 
> Since you mention the master branch, I ran the same build recipe on the
> master branch. Results (compared with the release/7.1 branch):
> 
> 2) On macOS 11..13, there is a compilation error:
> I would guess that the cause is a missing '#include <locale.h>'.

It seems like it is xlocale.h on macOS.  It should be fixed by this
commit:

https://git.savannah.gnu.org/cgit/texinfo.git/commit/?id=5efd627915bd05d4c265f111991d371e0b5cfd27

> 3) The OpenBSD build fails like this:
> -------------------------------------------------------------------------------
> Making all in man
> Updating man page install-info.1
> cd ..//install-info && make ginstall-info
> `ginstall-info' is up to date.
> /bin/sh 
> /home/runner/work/ci-check/ci-check/texinfo-7.1.90-20240610/build-aux/missing 
> help2man --name="update info/dir entries" -I ../../man/ginstall-info.h2m 
> ..//install-info/ginstall-info -o install-info.1
> /home/runner/work/ci-check/ci-check/texinfo-7.1.90-20240610/build-aux/missing[81]:
>  help2man: not found
> WARNING: 'help2man' is missing on your system.
>          You should only need it if you modified a dependency of a man page.
>          You may want to install the GNU Help2man package:
>          <https://www.gnu.org/software/help2man/>
> *** Error 127 in man (Makefile:2112 'install-info.1': @name="update info/dir 
> entries" && echo "Updating man page install-info.1"  && exe_ext...)
> *** Error 1 in . (Makefile:1947 'all-recursive': @fail=;  if 
> (target_option=k; case ${target_option-} in  ?) ;;  *) echo 
> "am__make_running_w...)
> *** Error 2 in 
> /home/runner/work/ci-check/ci-check/texinfo-7.1.90-20240610/build 
> (Makefile:1861 'all')
> -------------------------------------------------------------------------------
> 
> help2man is meant to be needed when building from git, according to
> README-hacking. But here, I'm building from a tarball. Why is help2man
> needed in this situation?

It should not be needed.  I have no clear idea on what could be going on
here.  Could you please check the timestamps of
install-info/install-info.c and man/install-info.1

When I check on my debian testing after unpacking, it is
$ ls -l install-info/install-info.c man/install-info.1 
-rw-rw-r-- 1 dumas dumas 93044 juin   8 10:56 install-info/install-info.c
-rw-rw-r-- 1 dumas dumas  4359 juin   8 11:21 man/install-info.1

> 4) The Solaris 11 OmniOS build fails due to a test failure:
> 
> FAIL: test_scripts/encoded_non_ascii_test_epub.sh
> 
> Detailed log:
> -------------------------------------------------------------------------------
> 
> FAIL: test_scripts/encoded_non_ascii_test_epub.sh
> =================================================
> 
> D: encoded/diffs/non_ascii_test_epub.diff (printed below)
> Common subdirectories: 
> ../../../tp/tests/encoded/res_parser/non_ascii_test_epub/osé_utf8_epub_package
>  and encoded/out_parser/non_ascii_test_epub/osé_utf8_epub_package
> Common subdirectories: 
> ../../../tp/tests/encoded/res_parser/non_ascii_test_epub/osé_utf8_epub_package/EPUB
>  and encoded/out_parser/non_ascii_test_epub/osé_utf8_epub_package/EPUB
> Common subdirectories: 
> ../../../tp/tests/encoded/res_parser/non_ascii_test_epub/osé_utf8_epub_package/META-INF
>  and encoded/out_parser/non_ascii_test_epub/osé_utf8_epub_package/META-INF
> Only in 
> ../../../tp/tests/encoded/res_parser/non_ascii_test_epub/osé_utf8_epub_package/EPUB:
>  0
> Common subdirectories: 
> ../../../tp/tests/encoded/res_parser/non_ascii_test_epub/osé_utf8_epub_package/EPUB/images
>  and encoded/out_parser/non_ascii_test_epub/osé_utf8_epub_package/EPUB/images
> Common subdirectories: 
> ../../../tp/tests/encoded/res_parser/non_ascii_test_epub/osé_utf8_epub_package/EPUB/xhtml
>  and encoded/out_parser/non_ascii_test_epub/osé_utf8_epub_package/EPUB/xhtml
> D: encoded/diffs/non_ascii_test_epub.diff (printed above)
> testdir: encoded
> driving_file: ../../../tp/tests/encoded/list-of-tests
> made result dir: ../../../tp/tests/encoded/res_parser/
> 
> doing test non_ascii_test_epub, src_file 
> ../../../tp/tests/encoded/osé_utf8.texi
> format_option: 
> texi2any.pl non_ascii_test_epub -> encoded/out_parser/non_ascii_test_epub
>  /usr/bin/perl -w ../../../tp/tests/..//texi2any.pl  --force --conf-dir 
> ../../../tp/tests/../t/init/ --conf-dir ../../../tp/tests/../init --conf-dir 
> ../../../tp/tests/../ext -I ../../../tp/tests/encoded -I encoded/ -I 
> ../../../tp/tests/ -I . -I built_input --error-limit=1000 -c TEST=1  --output 
> encoded/out_parser/non_ascii_test_epub/ -D 'needcommandlineunicode Need 
> command-line unicode' -c COMMAND_LINE_ENCODING=UTF-8 -c 
> MESSAGE_ENCODING=UTF-8 -c OUTPUT_FILE_NAME_ENCODING=UTF-8 --init epub3.pm -c 
> 'EPUB_CREATE_CONTAINER_FILE 0' ../../../tp/tests/encoded/osé_utf8.texi > 
> encoded/out_parser/non_ascii_test_epub/osé_utf8.1 
> 2>encoded/out_parser/non_ascii_test_epub/osé_utf8.2
> 
> all done, exiting with status 1
> FAIL test_scripts/encoded_non_ascii_test_epub.sh (exit status: 1)
> -------------------------------------------------------------------------------

This is very strange.  I do not see a file named 0 in the tarball in
tp/tests/encoded/res_parser/non_ascii_test_epub/osé_utf8_epub_package/EPUB/

The texi2any call should not change anything in the res_parser/.....
directories where the references for the test results are.  So my wild
guess is that there is something in the tp/tests/run_parser_all.sh shell
script driving the tests and comparisons with references that leads to
this file with name "0" to be created in the
tp/tests/encoded/res_parser/non_ascii_test_epub/osé_utf8_epub_package/EPUB/
directory, probably something not portable.  Maybe in the diff call,
maybe some use of $?, hard to know.  Also it is not clear why only this
test fails that way, other should fail too.

This test seems to pass on the opencsw solaris 11 CI:
https://buildfarm.opencsw.org/buildbot/waterfall?category=texinfo


Thanks for the reports!

-- 
Pat



reply via email to

[Prev in Thread] Current Thread [Next in Thread]