[Pkg-gauche-devel] gauche 0.9.4

[NIIBE Yutaka]

On 08/30/2014 04:47 AM, Jens Thiele wrote:
>> We have FTBFS for sparc, ppc64el, hurd-i386, among others.
>>
>> I'm considering to try ppc64el, but I don't know if I can support sparc
>> or hurd-i386.
> 
> for sparc maybe this is enough (didn't test wether the patch
> applies/works):
> 
> https://github.com/ivmai/bdwgc/commit/e2bf29b7801357c1ad28c2948a3f88c655dd722a
> see also:
> https://github.com/ivmai/bdwgc/issues/37

Thank you.  But, this is irrelevant.  This is already included in the
libgc in Gauche 0.9.4.

I've fixed build for sparc.  The problem is IA-64 and Sparc specific,
which requires files under gc/src.  sparc build is now successful.

I've also fixed some issues of ppc64el, but we encountered libgc test
failure on buildd machine.  I've tested ppc64el porterbox, but it is
not reproducible and it worked well there.  So, I believe that it's
the issue of ppc64el port itself.

We encountered failure for powerpc, but I believe this is the issue of
buildd machine for powerpc, as 0.9.4-2 was successful.

We encountered failure for s390x, but I believe this is the issue of
buildd machine for s390x, as 0.9.4-2 was successful.  But we should
care the testsuite of Gauche is somehow heavy for small resource
machines (*).

We encountered failure for hurd-i386, but I believe this is the issue
of Hurd itself.  The failure is at:
-----------------------
make[2]: Entering directory '/«PKGBUILDDIR»/src'
Testing math.* modules ...                                       passed.
Testing data.* modules ...                                       passed.
Testing optimizer ...                                            passed.
Testing control ...
/bin/bash: line 1:  8064 Illegal instruction
GAUCHE_TEST_RECORD_FILE=../test.record top_srcdir=.. ./gosh -ftest
-I../test $testfile >> test.log
-----------------------

Perhaps, I will try to figure out what's going on here, using
porterbox of hurd-i386, when I will have enough time, but priority is
very low for that.

We encountered failure for hppa, I believe this is the issue of
hppa buildd enviroment where 'automake' reefers automake-1.14.
There will be work around to invoke automake-1.11 specifically.

We encountered failure for sh4, I believe this is the issue of
small resource problem like the one of s390x.  --> (*)

We encountered failure for x32.  This is upstream (Gauche) problem,
I don't know it's worth to support.
--