[Debian-med-packaging] packaging for camitk 3.3.1 ready

[Andreas Tille]

Hi Emmanuel,

On Fri, Apr 25, 2014 at 10:09:52PM +0200, Emmanuel Promayon wrote:
> Hello,
> 
> So for the time being, I just removed test #9 and #15 in the svn
> (although I am not too sure the remaining #6 will work everywhere
> else !).

It seems not to work here (amd64):

6:   application-testcomponentsvtkimage3:
6:   /tmp/buildd/camitk-3.3.1/camitk-build/Testing/Temporary/application-testcomponentsvtkimage3/brain.mha
6:   does not match
6:   /tmp/buildd/camitk-3.3.1/camitk-build/share/camitk-3.3/testdata/ctestdata/brain.mha
6: 
6: 
2/2 Test #6: application-testcomponentsvtkimage3 ...***Failed    0.10 sec

50% tests passed, 1 tests failed out of 2

Label Time Summary:
component-msh         =   0.82 sec
component-vtkimage    =   0.10 sec

Total Test time (real) =   1.07 sec

The following tests FAILED:
          6 - application-testcomponentsvtkimage3 (Failed)


> The next upstream release (probably versioned 3.4) will have more
> relevant unit tests. We also plan to add more autopkgtests. Version
> 3.4 should be ready before the jessie freeze you mentioned in
> another thread so that this package should be "stronger" by then.

Fine.
 
> >>I tested the packaging on a sid pbuilder chroot and ran the
> >>debian/tests with adt-run on a sid LXC environment. Everything seems
> >>to work.
> >
> >Hmmm, strange that my pbuilder does not seem to work. :-(
> 
> Second part about this problem:
> Yes, I agree about the strangeness factor!
> 
> I created a pbuilder environment using this command:
> pdebuild --architecture amd64 --buildresult /tmp --pbuilderroot
> "sudo DIST=sid ARCH=amd64"

I admit I do not remember how I created mine but I think I left
everything as default (which seems the case also for you).

> More specifically I mainly followed the instructions on this web
> page [1] (I just had to double the size of tmpfs in order to compile
> the package, as it uses a lot of ITK-based template code which need
> quite a lot of memory).

I also needed to free some space but this was at build time and I do not
think that it is influencing the failed test.

> Btw, the C10shell is great as it gets you a
> shell if the build fails.

Yup, I'm using this but I have no idea how to profit from this in this
specific case since I have no good idea how to proceed from this failed
test.
 
> I would be interested to try to reproduce the problem using the same
> type/configuration of pbuilder as yours (although as I mentioned in
> the first part, some unit tests might not fail the same way as I am
> not sure they are yet "universal"). If you have any step by step
> that would help me to create the same pbuilder (or URL containing a
> step by step), that would be great. I noticed the "COW" keyword, is
> it a cowbuilder? What would be the difference between the pbuilder I
> used and yours?

Cowbuilder is somehow "the faster pbuilder" since it uses

   _c_opy _o_n _w_rite

I do not think that it makes any difference regardint the test.  I
think I created my environment using simply

   cowbuilder --create

and I'm calling

   /usr/bin/pdebuild --pbuilder cowbuilder

to build a package.

> Sorry in advance for my (blissed) ignorance and thank you again for
> your help.

Kind regards

         Andreas.

-- 
http://fam-tille.de