[Pkg-pascal-devel] CGE fails on two archs where it passed in the past

[Michalis Kamburelis]

2018-02-12 22:14 GMT+01:00 Paul Gevers <elbrus at debian.org>:
> Hi Michalis,
>
> CGE 6.4 fails it's tests on m68k and powerpc. CGE 6.2 passed on those
> archs. Maybe you want to have a look? (ppc64 fails for the same reason,
> so maybe you get a new arch for free if we fix it).
>
> https://buildd.debian.org/status/package.php?p=castle-game-engine
>

Thanks for the notice!

Looking briefly, I do not know why these tests (started) failing. We
had a lot of changes (both to the engine code as well as tests code),
so certainly something could happen. But they work on "normal"
platforms where I test (i386 and amd64), and I don't see there any
code that should behave differently depending on architecture (well,
except TestHash, it's very possible our StringToHash is accidentally
endianess-dependent).

I would need access to a porter box to see what's going on... Let's
first wait for me to deal with the porter boxes I have (arm64, mips,
mipsel) :) I hope to have time to deal with this by the end of
February.

Out of curiosity: These platforms were working with CGE 6.2, but when
CGE 6.2 was first uploaded to Debian, FPC version was older (not
3.0.4). Were these tests rerun when you uploaded new FPC 3.0.4? That
is, are we sure that these errors first appeared because of new CGE
6.4 version, and they were not introduced by new FPC 3.0.4 version?

Regards,
Michalis