Re: Race condition with 'make test'?

[Jean Abou Samra]

Le 16/11/2021 à 22:55, Aaron Hill a écrit :
> On 2021-11-16 12:54 pm, Jean Abou Samra wrote:
> > I cannot reproduce (I tried both normal builds and
> > with ./autogen.sh --enable-cairo-backend).
>
> In case it matters, the only flag I used with configure was 
> --disable-optimising.
>
>
> > - What do you get when running the grepping shell
> >   commands advised at the top of the make test output?
>
> $ grep -L systems.texi out/lybook-testdb/*/*log | sed s/log/ly/g | 
> xargs grep -H sourcefilename
> (Returned nothing?!)


OK, on second thought, it isn't surprising: lilypond-book
tests work a bit differently, so I think you would have
to remove the leading '*/' in '*/*log'. Or what if after
a failure you open the file
build/out/lybook-testdb/snippet-names-[the-sha-printed].log
directly?


> $ find out/lybook-testdb -name "*.log" | xargs grep -H 
> "\(error\|warning\)"
> out/lybook-testdb/01/lily-f9b407bf.log:Renaming input to: 
> `/home/seraku/sources/seraku24-lilypond/input/regression/markup-brace-warning.ly'
> out/lybook-testdb/0b/lily-db209039.log:/home/seraku/sources/seraku24-lilypond/input/regression/debug-property-callbacks.ly:21:3: 
> programming error: cyclic dependency: calculation-in-progress 
> encountered for #'stencil (NoteHead)
> [...]


So it doesn't find any error message in the relevant
tests. Must be an error of the kind that does not warn.


> > - Does it reproduce on a slightly older version, say,
> >   release/2.23.4-1? If so, it might be related to
> >   Han-Wen's recent lilypond-book infrastructure changes.
>
> Yes, I get the same errors against release/2.23.4-1 (9e187fb3). I will 
> try earlier tags and then see about bisecting.

Thanks.

Jean