Stefan writes:
Hi Stefan,
>> Yes, and there's a fix for that in mes-0.27.1
>
> OK, let me check with that version …
>
> It compiles with Mes 0.27.1 and NYACC 2.02.4!
Good
> But all mescc tests fail.
>
> starting phase `check'
> FAIL: lib/tests/scaffold/t.c
> FAIL: lib/tests/scaffold/01-return-0.c
> FAIL: lib/tests/scaffold/02-return-1.c
> FAIL: lib/tests/scaffold/03-call.c
> FAIL: lib/tests/scaffold/04-call-0.c
> ^C
How weird, have you checked
mescc-lib/lib/tests/scaffold/01-return-0.log
> Same for NYACC 2.02.3, 2.02.2, 2.00.6, 1.09.5, 1.09.02. Using Mes
> 0.27.1 and NYACC 1.09.1 the tests are working.
Hmm, interesting.
> Is it possible that before a release all tests of mescc are usually
> executed with guile and not mes? I suspect some problem in mes
> compared to guile, which get triggered with NYACC code above 1.09.1.
No, and with each release also a guix package is built for x86_64-linux
(which uses mes+mescc). Currently, mes in guix uses nyacc-2.02.2, but I
can confirm that 2.02.4 also works (for me).
> On the other side my patches may be bound to certain constructs form
> NYACC 1.09.1, which got changed in 1.09.2.
Ah, sure. That could be.
> But the test 01-return-0.c is a real basic one.
Indeed, given that everything was already compiled, I rather suspect a
linkage failure.
> I took the time to run build-aux/check-mes.sh and
> build-aux/check-boot.sh for the pair of Mes 0.27.1 and NYACC 2.02.4:
> They all pass. And all 188 tests from build-aux/check-mescc.sh fail.
That's so strange! I'm using m2-planet-1.12.0 and mescc-tools-1.7.0.
I've rebased wip-bootstrap-x86_64 onto version-0.27.1 and that also
works for me.
Greetings,
Janneke
--
Janneke Nieuwenhuizen <[email protected]> | GNU LilyPond https://LilyPond.org
Freelance IT https://www.JoyOfSource.com | Avatar® https://AvatarAcademy.com
