all-modules: Fix warnings in --version output.

2024-04-03 Thread Collin Funk
Running this script: $ ./all-modules --version ./all-modules: line 44: func_gnulib_dir: command not found sed: can't read /ChangeLog: No such file or directory ./all-modules: line 65: /build-aux/mdate-sh: No such file or directory all-modules (GNU gnulib ) Copyright (C)

Re: gnulib-tool.sh: Don't emit an unnecessary './' path component.

2024-04-03 Thread Bruno Haible
Hi Collin, > --- /home/collin/.local/src/texinfo/tp/Texinfo/XS/gnulib/lib/Makefile.am > 2024-04-03 04:25:14.105131041 -0700 > +++ > /home/collin/.local/src/texinfo/tp/Texinfo/glpyemzDvH/gnulib/lib/Makefile.am > 2024-04-03 04:25:02.276104588 -0700 > @@ -318,7 +318,7 @@ > ## begin

Re: gnulib-tool.py: Ignore pylint 'unidiomatic-typecheck' warnings.

2024-04-03 Thread Bruno Haible
Hi Collin, > I'm trying to fix some easy things to make warnings more meaningful. > Here is two patches for that. > > The first adds an option to the pylintrc configuration file. Right now > the pylint output gets spammed with 'unidiomatic-typecheck' warnings. > This is because we use: > >

Re: gnulib-tool.py: Don't ignore the argument to --doc-base.

2024-04-03 Thread Bruno Haible
Hi Collin, > When building GNU PSPP I see this: > > --- /home/collin/.local/src/pspp/gl/m4/gnulib-cache.m42024-04-02 > 18:10:06.209522277 -0700 > +++ /home/collin/.local/src/glpyXBzSID/gl/m4/gnulib-cache.m4 2024-04-02 > 18:09:37.828467148 -0700 > @@ -31,7 +31,7 @@ > # --lib=libgl \

Re: declaration after label

2024-04-03 Thread Bruno Haible
Paul Eggert wrote: > > Did C23 change this or is it just a compiler extension? > > C23 changed this. In C23 you can put a label before a declaration, or > before "}". ... and GCC ≥ 11 allows it also when compiling with -std=gnu99 or -std=gnu11. [1] [1]

Re: declaration after label

2024-04-03 Thread Paul Eggert
On 4/3/24 15:30, Collin Funk wrote: Did C23 change this or is it just a compiler extension? C23 changed this. In C23 you can put a label before a declaration, or before "}".

declaration after label

2024-04-03 Thread Collin Funk
On 4/3/24 2:33 PM, Bruno Haible wrote: > It has a syntax error: With clang and with GCC versions < 11, it is invalid > to put a declaration after a label. See: Good catch. I'm assuming that ISO C doesn't allow declarations after labels, which would explain the null statement (;) fixing it. Did

Re: quotearg: shell escape issue with single quote

2024-04-03 Thread Pádraig Brady
On 03/04/2024 22:33, Bruno Haible wrote: Hi Pádraig, The attached should fix this issue. process_input: + bool pending_shell_escape_end = false; It has a syntax error: With clang and with GCC versions < 11, it is invalid to put a declaration after a label. See: ===

Re: quotearg: shell escape issue with single quote

2024-04-03 Thread Bruno Haible
Hi Pádraig, > The attached should fix this issue. > process_input: > + bool pending_shell_escape_end = false; It has a syntax error: With clang and with GCC versions < 11, it is invalid to put a declaration after a label. See: === foo.c = int main () { goto

Re: quotearg: shell escape issue with single quote

2024-04-03 Thread Pádraig Brady
On 31/03/2024 22:02, Bruno Haible wrote: Pádraig Brady wrote: If a string starts with a sequence that requires $'' quoting followed by a \' and then another charactrer requiring $'' quoting, the '$' at the start of the quoted string ends up missing: $ env printf '%q\n' $'\1\'\2'

gnulib-tool.sh: Don't emit an unnecessary './' path component.

2024-04-03 Thread Collin Funk
I'm continuing to go through GNU packages and testing them with GNULIB_TOOL_IMPL=sh+py. Texinfo doesn't use the bootstrap script but the README-hacking file is easy to follow: To update the gnulib files, get a checkout of gnulib in a separate directory, then run, say

gnulib-tool.py: Ignore pylint 'unidiomatic-typecheck' warnings.

2024-04-03 Thread Collin Funk
I'm trying to fix some easy things to make warnings more meaningful. Here is two patches for that. The first adds an option to the pylintrc configuration file. Right now the pylint output gets spammed with 'unidiomatic-typecheck' warnings. This is because we use: if type(var) is str: