Re: Building libhmsbeagle in pbuilder mixes up desktop environment (Was: backport of libreoffice-calc (5.1.2-3~bpo8+1) mixes up desktop environment)
Hi Rebecca,
On Thu, Jul 14, 2016 at 10:17:43PM +0100, Rebecca N. Palmer wrote:
> >Are you able to reproduce this (under an up to date Jessie + backports
> >system).
>
> No - the build (of master, i.e. 2.1.2+20160525-1) fails with
>
> /tmp/pocldFu22d/program.cl:3273:32: error: call to '_cl_ldexp' is ambiguous
> sum[pat][state] += ldexp(frexp(dRootPartials[u + delta * r],
> &expTmp), expTmp + (tmpFactor - maxScaleFactor)) * matrixProp[r];
> ^~~~~
> during the test suite, but it doesn't do anything to graphics.
Strange. I do not get this.
> >What else information might be possibly helpful?
>
> Have you done anything to your pbuilder setup (pbuilderrc, gbp.conf) to make
> it install/use beignet?
The only thing that might give a hint is
BINDMOUNTS="/dev/shm"
I did not take any special means for beignet and I think I should not
since autobuilders will not have this as well - or am I wrong here?
> A program in a chroot can only see the OpenCL ICDs
> installed inside the chroot, which with those build-dependencies defaults to
> pocl-opencl-icd.
I admit I personally have no idea about opencl. I followed an advise on
debian-mentors or debian-science list when setting these dependencies.
> Does the problem go away if you uninstall beignet-opencl-icd and reboot?
I guess so but currently I have no time to try. Last time deinstalling
beignet-opencl-icd helped avoiding the screen problems.
> Do sid builds also trigger it?
The recent upload was done on a testing machine (but other hardware as
well) and has not shown any strange effect.
Kind regards
Andreas.
--
http://fam-tille.de
Reply to: