checkbox-dev team mailing list archive

[Chris Gregan <chris.gregan@xxxxxxxxxxxxx>]

Just ran it on my vivid test machine. Seems we had similar experiences, but
I believe the apport pop ups are correctly prompted since my fails were
actually crashes. Either way, however, we would want to suppress that.

Chris

On Wed, Mar 11, 2015 at 2:07 PM, Zygmunt Krynicki <
zygmunt.krynicki@xxxxxxxxxxxxx> wrote:

> Hey
>
> I'd like to update you on the status of getting piglit to work.
>
> First of all, there is a version of piglit stuck in Debian NEW [1].
> It's been there for the past five months. I'm trying to understand
> what the problem is but so far I didn't get any answers.
>
> Second of all, I had a look at a repository with the packaging and
> tried to build that. This failed on some unresolved symbols for
> linking. Updating dependencies and rebasing that on latest upstream
> master and changing some dependencies gave me a working package
> (though it needs love, see below).
>
> Running packaged piglit is a little unfriendly (I had no prior
> knowledge of how to use it) and I resorted to using it through our
> jobs in the checkbox provider. That quickly made it evident that the
> command line interface has change and our wrapper around piglit also
> needs changes.
>
> I managed to run the new piglit (guessing how it works from
> backtraces) by running "piglit run shader /tmp/results". This is still
> ongoing (12K tests). Piglit itself is *huge*, the package has over 23K
> files in it.
>
> As a note, I'd like to say that piglit could become a provider for
> plainbox. There are numerous similarities between plainbox and
> piglit's own framework. Perhaps we could create a template that simply
> exposes all of the many thousands of piglits tests to plainbox
> directly. Apart from being an interesting experiment in making our
> core efficient I can also anticipate the problem of aggregation.
> Though perhaps graphics card vendors would like a report with 12,000
> detailed entries on shaders alone.
>
> I've pushed my (rough) repository to [2] and a pre-built package to
> [3] (though the package needs more work, it's not good quality at all
> yet). Please give them a go and tell me what you find (you will need
> vivid to install that package though). My own testing has found that
> you get annoying apport pop-ups whenever a test fails which should be
> disabled before we try to use it.
>
> [1] https://ftp-master.debian.org/new/piglit_0~git20140926-4efb025-1.html
> [2] http://git.zygoon.pl (this is hosted at my home so be considerable
> of the bandwidth)
> [3]
> https://www.dropbox.com/sh/crhzlzqyhnga5au/AADcBcwro05xI6R35qKdMgqFa?dl=0
>



-- 
Chris Gregan
Project Manager
Professional and Engineering Services
Canonical USA Inc.
cgregan@xxxxxxxxxxxxx
cgregan[irc.freenode.net]
W-781-761-9448

----
1024/8806032D
E70F 7391 6C78 9B9E 6461 1CC7 B168 E1E7 8806 032D