Control: severity -1 normal
pocl is back in testing
On 12/01/2022 20.33, Rebecca N. Palmer wrote:
Probably just Build-Depends: pocl-opencl-icd [amd64 arm64 armel armhf
i386 hurd-i386] and skip the tests on the other architectures. (Linking
uses libopencl1, not *-opencl-icd.)
I'll keep this
On Wed, 12 Jan 2022 19:33:11 + "Rebecca N. Palmer"
wrote:
Probably just Build-Depends: pocl-opencl-icd [amd64 arm64 armel armhf
i386 hurd-i386] and skip the tests on the other architectures. (Linking
uses libopencl1, not *-opencl-icd.)
As a slightly unrelated note, if that Build-Depends
On 2022-01-12 20:00, Rebecca N. Palmer wrote:
On 12/01/2022 12:50, Drew Parsons wrote:
mesa-opencl-icd is running and passing its own tests on these arches.
What is pyopencl doing in build and tests that's different from what
mesa-opencl-icd itself does?
Probably that pyopencl's tests actuall
On 2022-01-12 00:04, Rebecca N. Palmer wrote:
That's probably a good idea for runtime Depends, but *not*
Build-Depends and test-Depends: those are likely to be run on GPU-less
servers, where only CPU ICDs (i.e. pocl-opencl-icd) will work.
(Roughly: pocl-opencl-icd = CPUs, mesa-opencl-icd = AMD/A
Source: pyopencl
Version: 2021.2.9-1
Followup-For: Bug #1001545
Reading the history further, mesa-opencl-icd was replaced by
pocl-opencl-icd | opencl-icd for 2016.1+git20161003-2,
commit 3b0d193c,
"pocl-opencl-icd is the best default choice when there no other OpenCL
ICD installed as it uses CPU a
Source: pyopencl
Followup-For: Bug #1001545
It should be possible to close this bug soon once pocl 1.8-3 migrates
to testing.
If pocl continues to make trouble, a workaround could be to replace
Build-Depends: pocl-opencl-icd
with
Build-Depends: mesa-opencl-icd | pocl-opencl-icd | opencl-icd
Package: pyopencl
Tags: bookworm
Severity: serious
Justification: rc policy - "Packages must be buildable within the same
release"
pyopencl build-depends on pocl-opencl-icd which is no longer in testing.
7 matches
Mail list logo
