Re: [Bug-datamash] datamash 1.3 - Bug on ARM and PowerPC architectures

[Guillaume Gardet]

Hi,


Le 22/03/2018 à 18:54, Assaf Gordon a écrit :
> Hello,
>
> On Thu, Mar 22, 2018 at 02:46:01PM +0100, Guillaume Gardet wrote:
> > on openSUSE, we are building datamash for x86 without problem, but PowerPC and ARM (32 
> > and 64 bits), are failing on a single test: "tests/datamash-output-format.pl", 
> > here is the output log:
> >
> > **********************************************************************
> > a1...
> > datamash-output-format.pl: test a1: stdout mismatch, comparing a1.2 (expected) 
> > and a1.O (actual)
> > *** a1.2        Thu Mar 22 13:36:38 2018
> > --- a1.O        Thu Mar 22 13:36:38 2018
> > ***************
> > *** 1 ****
> > ! 0x8.000p-3
> > --- 1 ----
> > ! 0x1.000p+0
> > **********************************************************************
> Thank you for reporting this issue.
>
> This is a false positive, due to the fact the there could be multiple
> (valid) representation of 64bit double in hex-scientific format
> (i.e. printf("%a")).
>
> So it is safe to ignore this failure.

Ok.

> However,
> I just pushed two commits that disable this test
> (and fix another test).
>
> Available here:
> https://git.savannah.gnu.org/cgit/datamash.git/commit/?id=d446dba3
> https://git.savannah.gnu.org/cgit/datamash.git/commit/?id=f4871963

Thanks a lot. Those 2 patches fixed AArch64, armv7 and PPC64.

Best regards,

Guillaume

> And also attached as a patch to this message.
>
> regards,
>   - assaf