[Reproducible-builds] Question about build environments used for i386

[Holger Levsen]

On Thu, May 26, 2016 at 10:42:51AM +0200, Gert Wollny wrote:
> Hi, 
> 
> Am Donnerstag, den 26.05.2016, 08:27 +0000 schrieb Mattia Rizzolo:
> > On Thu, May 26, 2016 at 08:08:07AM +0200, Gert Wollny wrote:
> > > 
> > > looking at the non-reproducibility of dcmtk [1], I found that for
> > > some reason on i386 in one build cmake obtains "x86_64" as system
> > > processor  versus "i686" the in other build. 
> > That's because for i386 one build is done with a i386 kernel, and the
> > other with a amd64 kernel, but both with a i386 userland.
> I see. 
> 
> > > CMake supposedly uses "uname -p" to obtain this information and now
> > > I'm wondering about the build environment of the first build?
> > `uname -p` sounds like an ugly choice, also considering this:
> > mattia at chase ~ % uname -p
> > unknown
> Yeah, I've seen the same in my setup, so I will report a bug for cmake
> upstream. 
> 
> > > Unfortunately, the provided logs are only for amd64.
> > Not sure what are you referring to here.
> > For i386 there are both logs for build builds done on i386.
> Okay, now I've found them, the interface is a bit irritating though,
> because when opening the reproducible builds page amd64 is the default,
> and it eluded me that when clicking on one of the i386 builds the
> details in the pane above change. It would probably be better  to have
> the archs first listed and the according (changing) details below.
> 
> Thanks, 
> Gert 
> 
> 
> _______________________________________________
> Reproducible-builds mailing list
> Reproducible-builds at lists.alioth.debian.org
> http://lists.alioth.debian.org/cgi-bin/mailman/listinfo/reproducible-builds

-- 
cheers,
	Holger
-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 811 bytes
Desc: Digital signature
URL: <http://lists.alioth.debian.org/pipermail/reproducible-builds/attachments/20160526/21c44965/attachment.sig>