How to deal with failing 'make test-bseline'

[Thomas Morley]

Bringing some offlist-discussion back to the list because it may
interest more people

Am So., 9. Feb. 2020 um 12:41 Uhr schrieb Thomas Morley
<address@hidden>:
>
> Am So., 9. Feb. 2020 um 12:12 Uhr schrieb David Kastrup <address@hidden>:
> >
> > Dan Eble <address@hidden> writes:
>
> > > It would be nice if the test driver would not abort the whole
> > > regression test suite when one instance of lilypond exits with an
> > > error.
> >
> > Well, there was a reason you turned warnings into errors for this test:
> > making people notice.
>
> For me I wonder how you've got the error message:
>
> Am Sa., 8. Feb. 2020 um 17:31 Uhr schrieb David Kastrup <address@hidden>:
>
> > input/regression/display-lily-tests.ly:230:1: fatal error: Test unequal: 
> > BUG.
> > in  = \applyOutput Foo #(lambda (arg) (list))
> > out = \applyOutput Foo ##f
> >
> >
> > \test ##[ \applyOutput Foo #(lambda (arg) (list)) #]
> >
>
> As written here
> https://lists.gnu.org/archive/html/lilypond-devel/2019-10/msg00236.html
> every 'make test-baseline' I tried failed.
> And I never got a meaningful error: usually a log-file with the
> lily-version and nothing else.
> As one can read in the linked thread, I likely followed the wrong track then,
>
> Seeing such an error message like you, I may have been able to tackle
> the problem months ago.


Am So., 9. Feb. 2020 um 12:44 Uhr schrieb David Kastrup <address@hidden>:

> The output of "make test" starts with a line "for tracing crashes, run
> the following command" followed by some obscure grep recipe.  I have no
> idea what this recipe does, but I have no idea how people have a chance
> to find problems without using it.
>
> In this case, it pinpointed the problematic file, and running that
> manually gave the error message in question.

Ok.
I started from scratch (without #5743), thus 'make test-baseline' _should_ fail.
And it does.

The message
Please see 
/home/hermann/lilypond-git/build/out/lybook-testdb/snippet-names-4cf679323625678e6dfeeb9e33801c5e.log
points to a log-file containing nothing else than the version.

Doing
~/lilypond-git/build (master)$ grep sourcefilename `grep -L
systems.texi out/lybook-testdb/*/*log|sed s/log/ly/g`
returns:
\sourcefilename "/home/hermann/lilypond-git/input/regression/slur-rest.ly"
Compiling input/regression/slur-rest.ly works like a charm, though.

So what does this mean?
Am I missing something?
Or is the whole repo corrupted and I should restart with a fesh clone?
(At least this is not the repo I push from...)

Cheers,
  Harm