| From: | "Jonathan Gonzalez V(dot)" <jonathan(at)abdiel(dot)eu> |
|---|---|
| To: | Peter Eisentraut <peter(at)eisentraut(dot)org> |
| Cc: | pgsql-hackers <pgsql-hackers(at)postgresql(dot)org> |
| Subject: | Re: Avoid duplicate file references in backend .pot file |
| Date: | 2026-11-03 11:17:40 |
| Message-ID: | 87pkwmduzv.fsf@abdiel.eu |
| Views: | Whole Thread | Raw Message | Download mbox | Resend email |
| Thread: | |
| Lists: | pgsql-hackers |
Hello!
Peter Eisentraut <peter(at)eisentraut(dot)org> writes:
> The list of files for the backend's message catalog is collected with find,
> which also lists symlinks, namely the generated files that are linked from
> src/include/utils/ to src/backend/utils/ (such as guc_tables.inc.c) and the
> port/pg_sema.c and port/pg_shmem.c links created by configure. Since the link
> targets are found under their real names as well, xgettext processed those files
> twice, resulting in duplicate "#:" references like
>
> #: ../include/utils/guc_tables.inc.c:410 utils/guc_tables.inc.c:410
>
> This is an old problem, but the introduction of guc_tables.inc.c in PG19 has
> made it much more noticeable.
>
> I propose the attached patch to fix this by restricting find to regular files.
> gettext-files: generated-parser-sources generated-headers
> - find $(srcdir) $(srcdir)/../common $(srcdir)/../port $(srcdir)/../include/ \( -name '*.c' -o -name "proctypelist.h" \) -print | LC_ALL=C sort >$@
> + find $(srcdir) $(srcdir)/../common $(srcdir)/../port $(srcdir)/../include/ -type f \( -name '*.c' -o -name "proctypelist.h" \) -print | LC_ALL=C sort >$@
This make a lot of sense to me, I've tested and found that
./../include/utils/pgstat_wait_event.c
./../include/utils/wait_event_funcs_data.c
both were also symlinks.
This case totally escaped to my tests the last I checked that file! Work
as expected!
Regards,
--
Jonathan Gonzalez V.
EDB
https://www.enterprisedb.com
| From | Date | Subject | |
|---|---|---|---|
| Next Message | Jakub Wartak | 2026-11-03 11:31:30 | Re: [Bug Report + Patch] File descriptor leak when io_method=io_uring |
| Previous Message | Virender Singla | 2026-11-03 11:17:01 | Re: Allow pg_read_all_stats to read replication origin status |