Re: gfortran transition update
Greetings, again!
Just spent a few minutes reviewing our project machines availability
for the various platforms. Many are still locked down from the
breakin many many months ago. Requests to debian-admin go
unanswered. I have several previously portable packages which are now
held up for months due in part to lack of access to a proper debugging
environment. It appears as if this situation is not likely to be
remedied soon. (In particular, alpha, arm, and sparc appear to be the
most egregious offenders.)
I like everyone else have limited time to devote to Debian, but of
course value to the project and especially its portability dearly.
Nevertheless, it appears that it might be wise for me, and perhaps for
others, to adopt one of the following procedures for their packages:
1) mark for i386 only, and let others be resonponsible for porting
2) upload binaries built under a cross compiling environment, assuming
an acceptable one exists.
I think that all bugs should be marked as minor if they are arch
specific and if no machine is provided for at least a contiguous three
months to address the issue.
Comments?
Take care,
Riku Voipio <riku.voipio@iki.fi> writes:
> Hi,
>
> Firstly, I've collected all (well, most) information bits to:
>
> http://wiki.debian.org/GfortranTransition
>
> Feel free to update/edit the page. Most importantly, we need to get
> the packages that need a library rename done uploaded to experimental
> soon.
>
> Second refblas3 testsuite on gfortran failed again on many architectures:,
>
> http://experimental.debian.net/build.php?pkg=refblas3
>
> Amd64 and mips fail with the following tests:
>
> Test of subprogram number 1 CBLAS_SDOT
> FAIL
>
> CASE N INCX INCY MODE I COMP(I) TRUE(I) DIFFERENCE SIZE(I)
>
> 1 1 1 1 9999 1 0.20000000E+01 0.30000001E+00 0.1700E+01 0.3000E+00
> 1 2 1 1 9999 1 -0.20000000E+01 0.20999999E+00 -0.2210E+01 0.1600E+01
> 1 4 1 1 9999 1 0.20000000E+01 0.62000000E+00 0.1380E+01 0.3200E+01
> 1 1 2 -2 9999 1 0.20000000E+01 0.30000001E+00 0.1700E+01 0.3000E+00
> 1 2 2 -2 9999 1 0.00000000E+00 -0.70000000E-01 0.7000E-01 0.1600E+01
> 1 4 2 -2 9999 1 0.36893488E+20 0.85000002E+00 0.3689E+20 0.3200E+01
> 1 1 -2 1 9999 1 0.20000000E+01 0.30000001E+00 0.1700E+01 0.3000E+00
> 1 2 -2 1 9999 1 0.10842022E-18 -0.79000002E+00 0.7900E+00 0.1600E+01
> 1 4 -2 1 9999 1 0.36893488E+20 -0.74000001E+00 0.3689E+20 0.3200E+01
> 1 1 -1 -2 9999 1 0.20000000E+01 0.30000001E+00 0.1700E+01 0.3000E+00
> 1 2 -1 -2 9999 1 0.36893488E+20 0.33000001E+00 0.3689E+20 0.1600E+01
> 1 4 -1 -2 9999 1 0.00000000E+00 0.12700000E+01 -0.1270E+01 0.3200E+01
>
> The E+20 are worrying.
>
> mipsel, sparc, s390 and hppa fail with slightly different results:
>
> Test of subprogram number 1 CBLAS_SDOT
> FAIL
>
> CASE N INCX INCY MODE I COMP(I) TRUE(I) DIFFERENCE SIZE(I)
>
> 1 1 1 1 9999 1 0.16500000E+01 0.30000001E+00 0.1350E+01 0.3000E+00
> 1 2 1 1 9999 1 0.15849999E+01 0.20999999E+00 0.1375E+01 0.1600E+01
> 1 4 1 1 9999 1 0.17800000E+01 0.62000000E+00 0.1160E+01 0.3200E+01
> 1 1 2 -2 9999 1 0.16500000E+01 0.30000001E+00 0.1350E+01 0.3000E+00
> 1 2 2 -2 9999 1 -0.13900000E+01 -0.70000000E-01 -0.1320E+01 0.1600E+01
> 1 4 2 -2 9999 1 0.18375000E+01 0.85000002E+00 0.9875E+00 0.3200E+01
> 1 1 -2 1 9999 1 0.16500000E+01 0.30000001E+00 0.1350E+01 0.3000E+00
> 1 2 -2 1 9999 1 -0.18225000E+01 -0.79000002E+00 -0.1033E+01 0.1600E+01
> 1 4 -2 1 9999 1 -0.18099999E+01 -0.74000001E+00 -0.1070E+01 0.3200E+01
> 1 1 -1 -2 9999 1 0.16500000E+01 0.30000001E+00 0.1350E+01 0.3000E+00
> 1 2 -1 -2 9999 1 0.16650000E+01 0.33000001E+00 0.1335E+01 0.1600E+01
> 1 4 -1 -2 9999 1 0.19087499E+01 0.12700000E+01 0.6387E+00 0.3200E+01
>
> These could be some kind of rounding disparity.
>
>
> --
> "rm -rf" only sounds scary if you don't have backups
>
>
>
--
Camm Maguire camm@enhanced.com
==========================================================================
"The earth is but one country, and mankind its citizens." -- Baha'u'llah
Reply to: