[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: [Chicken-hackers] [PATCH] Several fixes to make CHICKEN work on Sola
From: |
Peter Bex |
Subject: |
Re: [Chicken-hackers] [PATCH] Several fixes to make CHICKEN work on Solaris again |
Date: |
Fri, 21 Feb 2014 20:06:10 +0100 |
User-agent: |
Mutt/1.4.2.3i |
On Fri, Feb 21, 2014 at 01:18:11AM +0000, Mario Domenech Goulart wrote:
> Thanks, Peter. I've pushed them all after testing them on the following
> platforms:
>
> OS HW Arch C compiler
> ------------------------+-----------+----------
> Linux x86-64 GCC
> Linux x86-64 clang
> FreeBSD 10 x86-64 clang
> Windows XP (mingw) x86 GCC
> Windows XP (cygwin) x86 GCC
> Windows XP (mingw-msys) x86 GCC
> Haiku x86 GCC
Wow, fantastic! Thanks for this very thorough testing!
> I faced only a small issue: "make check" hangs on Haiku:
>
> ======================================== runtime tests ...
> many arguments supported.
> testing 'apply' with 0..2048 (maximum apply argument count)...
> If this segfaults on x86-64, try updating GCC (4.5 has a code-generation bug):
> invoking directly with 0..50...
> invoking directly with 1998..2048 (maximum interpreted direct argument
> count)...
> testing 'apply' can detect calls of too many arguments...
> testing direct invocation can detect calls of too many arguments...
> ../chicken 'apply-test.scm' -output-file 'a.c' -verbose -include-path ..
> 'gcc' 'a.c' -o 'a.o' -c -fno-strict-aliasing -fwrapv -DHAVE_CHICKEN_CONFIG_H
> -DC_ENABLE_PTABLES -Os -fomit-frame-pointer -I..
> -I"/Blank_BFS/local/chicken-master/include/chicken"
>
> It seems that gcc gets in an endless (?) loop or something that takes a
> very long time hogging the CPU. If I comment lines 145 and 146 in
> runtests.sh, "make check" finishes without problems.
I know this happens on some versions of GCC, and I guess it's especially
noticable on a slow VM. But if you wait long enough, it will eventually
finish. I think it has some trouble with the (foo 1 2 3 4 ... 2048) call
which we generate through a macro and translates directly to the C
code foo(1, 2, 3, 4, ... 2048); :)
> Despite that issue, which I believe to be GCC's and not related to your
> patches, I could install pastiche and run awful on Haiku.
I agree, it's not too bad.
Cheers,
Peter
--
http://www.more-magic.net