Rare fails of bibtex8 and/or bibtexu test

Dr. Werner Fink werner at suse.de
Thu Apr 13 08:18:02 CEST 2023

On 2023/04/12 11:59:33 -0600, Karl Berry wrote:
>     I face rare and random fails of tests/bibtex8.test and/or
>     tests/bibtexu.test
> Maybe it's a timing/parallelism/caching/something problem.
> Are you running with make -j?

Yep our OBS aka autobuild system uses VMs with different numbers of
CPUs mostly 8 and therefore -j8 ... and this leads to random order
of the bibtex[u8] tests.

Meanwhile I've a patch

--- texlive-20230311-source/texk/bibtex-x/Makefile.in
+++ texlive-20230311-source/texk/bibtex-x/Makefile.in   2023-04-12 13:47:24.619293127 +0000
@@ -13,7 +13,7 @@
 VPATH = @srcdir@
 am__is_gnu_make = { \

with this the order of the tests are repeatable for every run
and every architecture.

> I see bibtex8.test and bibtexu.test are using the same input and output
> filenames. I just changed bibtexu.test to change that (r66832). Hope
> that helps.

That also avoid this, indeed.

>     Should I increase string values of bibtex
> No. These test bibs are tiny. --thanks, karl.

Thanks a lot


  "Having a smoking section in a restaurant is like having
          a peeing section in a swimming pool." -- Edward Burr
-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 894 bytes
Desc: not available
URL: <https://tug.org/pipermail/tex-live/attachments/20230413/af5e08e4/attachment.sig>

More information about the tex-live mailing list.