RFS: gasic

classic Classic list List threaded Threaded
6 messages Options
Reply | Threaded
Open this post in threaded view
|

RFS: gasic

Pranav Ballaney
Hi,
I've added autopkgtests to gasic. Please review and sponsor.

Regards,
Pranav
Reply | Threaded
Open this post in threaded view
|

Re: RFS: gasic

Andreas Tille-5
Hi Pranav,

On Sun, May 10, 2020 at 02:58:32AM +0530, Pranav Ballaney wrote:
> I've added autopkgtests to gasic. Please review and sponsor.
> https://salsa.debian.org/med-team/gasic/

thank you for adding this test.  There is one cosmetic issue and a
problematic issue.  The cosmetic thing is that the additional data are
blowing up this quite small package a lot.  I do not think that we
should bloat user machines too much.  Thus it would be better to ship
the data in an extra binary package gasic-examples.

But here comes the harder part:  I get

autopkgtest [05:12:46]: test run-unit-test: [-----------------------
/tmp/autopkgtest.qJbg1v/tree/debian/tests/run-unit-test: line 23:   903 Segmentation fault      bowtie-build dwv.fasta dwv.fasta
autopkgtest [05:12:47]: test run-unit-test: -----------------------]

in my pbuilder chroot.  I remember I once had a similar issue with
bowtie before.  Its a bit hard to seek in BTS and I have no idea
whether it makes sense to seek at all.  But it would be great if you
could reproduce this somehow and we could debug that issue - which
is probably an issue in bowtie - it just should not segfault.

Kind regards

      Andreas.

--
http://fam-tille.de

Reply | Threaded
Open this post in threaded view
|

Re: RFS: gasic

Pranav Ballaney
Hi Andreas, 
I've created another binary called gasic-examples and moved all examples there. 

I ran the tests again, and they all pass in my chroot. I also rebuilt the chroot in case there was something wrong there, and turns out they still pass. I also tried adding:
     bowtie-build dwv.fasta dwv.fasta
to the autopkgtests for the bowtie package, and all of those pass too. 

For the purpose of completing autopkgtests for gasic, I have tried to circumvent this issue by changing the mapper for gasic from bowtie to bwa, so bowtie will no longer be required for testing gasic.
I've also fixed a patch and added some more tests.

Please take a look - specially at the new binary (in case I've misplaced some file) - and if these tests work now, please sponsor. 

Regarding bowtie, I'm unable to reproduce the issue, so if you have any ideas that I can try, please let me know.

Regards,
Pranav


On Sun, May 10, 2020 at 10:53 AM Andreas Tille <[hidden email]> wrote:
Hi Pranav,

On Sun, May 10, 2020 at 02:58:32AM +0530, Pranav Ballaney wrote:
> I've added autopkgtests to gasic. Please review and sponsor.
> https://salsa.debian.org/med-team/gasic/

thank you for adding this test.  There is one cosmetic issue and a
problematic issue.  The cosmetic thing is that the additional data are
blowing up this quite small package a lot.  I do not think that we
should bloat user machines too much.  Thus it would be better to ship
the data in an extra binary package gasic-examples.

But here comes the harder part:  I get

autopkgtest [05:12:46]: test run-unit-test: [-----------------------
/tmp/autopkgtest.qJbg1v/tree/debian/tests/run-unit-test: line 23:   903 Segmentation fault      bowtie-build dwv.fasta dwv.fasta
autopkgtest [05:12:47]: test run-unit-test: -----------------------]

in my pbuilder chroot.  I remember I once had a similar issue with
bowtie before.  Its a bit hard to seek in BTS and I have no idea
whether it makes sense to seek at all.  But it would be great if you
could reproduce this somehow and we could debug that issue - which
is probably an issue in bowtie - it just should not segfault.

Kind regards

      Andreas.

--
http://fam-tille.de

Reply | Threaded
Open this post in threaded view
|

Re: RFS: gasic

Andreas Tille-5
Hi Pranav,

On Wed, May 13, 2020 at 02:47:11AM +0530, Pranav Ballaney wrote:
> Hi Andreas,
> I've created another binary called gasic-examples and moved all examples
> there.

I've checked this and did some slight changes.  Please note that all the
dependencies you copied from gasic package are not needed to the examples.
You can simply install the data files without any dependencies.

I've also moved the data files to /usr/share/doc/gasic/examples.
Finally the user wants to have examples for gasic and chances are better
that they will be found there.
 
> I ran the tests again, and they all pass in my chroot. I also rebuilt the
> chroot in case there was something wrong there, and turns out they still
> pass. I also tried adding:
>      bowtie-build dwv.fasta dwv.fasta
> to the autopkgtests for the bowtie package, and all of those pass too.

I confirm that I injected that call into the build of the bowtie package
where it succeeded.  Strangely enough the whole autopkgtest of bowtie
failed in my pbuilder chroot which I'm using as a hook in my build
environment.  So may be this environment has some issues since bowtie at
ci.debian.org is fine.
 
> For the purpose of completing autopkgtests for gasic, I have tried to
> circumvent this issue by changing the mapper for gasic from bowtie to bwa,
> so bowtie will no longer be required for testing gasic.

Hmmm, I admit I've read this to late.  May be you forgot to push?  But
don't mind about this.  The package needs to go to the new queue due to
the new binary package and needs another source only upload afterwards.
We can add your changes than.

> I've also fixed a patch and added some more tests.
>
> Please take a look - specially at the new binary (in case I've misplaced
> some file) - and if these tests work now, please sponsor.

I've uploaded now - please check whether you have uncommitted changes and
merge these for the future upload.
 
> Regarding bowtie, I'm unable to reproduce the issue, so if you have any
> ideas that I can try, please let me know.

If you are using cowbuilder you might like to copy

   /usr/share/doc/pbuilder/examples/B20autopkgtest

to

   /var/cache/pbuilder/hooks/B20autopkgtest

This will run the autopkgtest right inside the pbuilder chroot.  This is
what I'm using in all my builds with less than 1% of false positives
like in the bowtie case.  Probably having another hook in analogy to
C99_failed_build to stay in the autopkgtest root and fire up gdb would
uncover some issues.
 
Thank you for your work on this

     Andreas.

--
http://fam-tille.de

Reply | Threaded
Open this post in threaded view
|

Re: RFS: gasic

Pranav Ballaney
Hi Andreas,
Indeed, I did forget to push. I had been working on these tests for hours and when they finally worked I got too excited to send the email. 😂

Regarding bowtie: I use sbuild as of now, but I'll try making a chroot with pbuilder and running the tests.
With the recently pushed changes, bowtie will no longer be required to test gasic, so these tests should pass for you as well, now.

Regards,
Pranav


On Wed, 13 May, 2020, 12:59 PM Andreas Tille, <[hidden email]> wrote:
Hi Pranav,

On Wed, May 13, 2020 at 02:47:11AM +0530, Pranav Ballaney wrote:
> Hi Andreas,
> I've created another binary called gasic-examples and moved all examples
> there.

I've checked this and did some slight changes.  Please note that all the
dependencies you copied from gasic package are not needed to the examples.
You can simply install the data files without any dependencies.

I've also moved the data files to /usr/share/doc/gasic/examples.
Finally the user wants to have examples for gasic and chances are better
that they will be found there.

> I ran the tests again, and they all pass in my chroot. I also rebuilt the
> chroot in case there was something wrong there, and turns out they still
> pass. I also tried adding:
>      bowtie-build dwv.fasta dwv.fasta
> to the autopkgtests for the bowtie package, and all of those pass too.

I confirm that I injected that call into the build of the bowtie package
where it succeeded.  Strangely enough the whole autopkgtest of bowtie
failed in my pbuilder chroot which I'm using as a hook in my build
environment.  So may be this environment has some issues since bowtie at
ci.debian.org is fine.

> For the purpose of completing autopkgtests for gasic, I have tried to
> circumvent this issue by changing the mapper for gasic from bowtie to bwa,
> so bowtie will no longer be required for testing gasic.

Hmmm, I admit I've read this to late.  May be you forgot to push?  But
don't mind about this.  The package needs to go to the new queue due to
the new binary package and needs another source only upload afterwards.
We can add your changes than.

> I've also fixed a patch and added some more tests.
>
> Please take a look - specially at the new binary (in case I've misplaced
> some file) - and if these tests work now, please sponsor.

I've uploaded now - please check whether you have uncommitted changes and
merge these for the future upload.

> Regarding bowtie, I'm unable to reproduce the issue, so if you have any
> ideas that I can try, please let me know.

If you are using cowbuilder you might like to copy

   /usr/share/doc/pbuilder/examples/B20autopkgtest

to

   /var/cache/pbuilder/hooks/B20autopkgtest

This will run the autopkgtest right inside the pbuilder chroot.  This is
what I'm using in all my builds with less than 1% of false positives
like in the bowtie case.  Probably having another hook in analogy to
C99_failed_build to stay in the autopkgtest root and fire up gdb would
uncover some issues.

Thank you for your work on this

     Andreas.

--
http://fam-tille.de

Reply | Threaded
Open this post in threaded view
|

Re: RFS: gasic

Andreas Tille-5
Hi Pranav,

On Wed, May 13, 2020 at 10:21:16PM +0530, Pranav Ballaney wrote:
> Indeed, I did forget to push. I had been working on these tests for hours
> and when they finally worked I got too excited to send the email. 😂

I can confirm that you are by far not the first team member who forgot
to push. ;-)
 
> Regarding bowtie: I use sbuild as of now, but I'll try making a chroot with
> pbuilder and running the tests.
> With the recently pushed changes, bowtie will no longer be required to test
> gasic, so these tests should pass for you as well, now.

I admit I do not consider it a drawback if we are testing other packages
that way.  So *if* bowtie might have an issue it would be good to spot
it that way.  But I think bwa is the more frequently used tool anyway so
its sensible to use this one.

Kind regards

       Andreas.

--
http://fam-tille.de