Re: Failure in autopkgtest for loki (Was: Working tests seem to exist for epcr)

2020-02-21 Thread Andreas Tille
Hi Pranav,

as a general note:  On Debian lists it is assumed that people (who do
not say otherwise) are subscribed to the list.  Thus in responses of
some mail the poster is usually not CCed.  In mutt exists the feature
"List-Reply" - no idea about your client.  I will follow this policy now
and just write to the list.  In our small Debian Med community its
kindly tolerated if you put the original poster in To/CC as well so you
if it might be an extra burden for you to edit To/CC just keep on
following your habit.  I'm just explaining why I'm not CCing you
personally any more (usually) and this is also a warning for you in case
you might once sent to other Debian lists on one hand people might
respond annoyed if they get e-mails twice and on the other hand you
might missing any response since you might not be subscribed to that
specific list (posting policy is that everybody can post).  In this case
I'd recommend reading the actual web archive.

On Fri, Feb 21, 2020 at 12:42:10PM +0530, Pranav Ballaney wrote:
> Hi Andreas,
> Thanks for taking the time out to look into my work.

You are perfectly welcome.  On one hand I see this as my duty as a
(pontential) GSoC mentor.  On the other hand your level of contribution
shows that you are really dedicated to the task and I'd love to support
this.
 
> On Fri, Feb 21, 2020 at 1:54 AM Andreas Tille  wrote:
> 
> > How are you running the test at your site?
> 
> I run the tests inside a chroot I made using sbuild-debian-developer-setup
> [1], and everything seems to be fine there as well.

In principle an sbuild setup - provided that it also has something like
a hook that is triggering the autopkgtest running in a chroot - should
lead to the same results.

> From the output, it seems that loki is trying to create a file called
> loki.dat, but failing.
> Just off the top of my head, could it be an issue with permissions? Does
> the chroot allow programs to create files?

Well, the script is copying files to that location - so file creation
permissions should not be an issue, IMHO.  I also checked the code line
in the source file that is throwing the error which looks not really
suspicious to me.  But I only had a quick look.  Its pretty old code so
we need to expect things that are not working as expected these days.

> I will try building a pbuilder environment on my machine as well, and look
> into this at the earliest.

Just take your time.  In case you might be blocked it can help if you
look into another package.  Liubov had announced to possibly have a look
at the weekend and sometimes it helps to have an additional pair of
eyeballs.  Sorry, for my "feature" to always stumble about issue. ;-)
 
> > Besides this issue you are perfectly on the right track and I repeat
> > that I'm absolutely happy about your performance right from the
> > beginning.
>
> Thank you so much!

Again, you are really welcome

   Andreas.

-- 
http://fam-tille.de



Re: Failure in autopkgtest for loki (Was: Working tests seem to exist for epcr)

2020-02-21 Thread Pranav Ballaney
Hi Andreas,

On Fri, Feb 21, 2020 at 3:57 PM Andreas Tille  wrote:

> Hi Pranav,
>
> as a general note:  On Debian lists it is assumed that people (who do
> not say otherwise) are subscribed to the list.  Thus in responses of
> some mail the poster is usually not CCed.  In mutt exists the feature
> "List-Reply" - no idea about your client.  I will follow this policy now
> and just write to the list.  In our small Debian Med community its
> kindly tolerated if you put the original poster in To/CC as well so you
> if it might be an extra burden for you to edit To/CC just keep on
> following your habit.  I'm just explaining why I'm not CCing you
> personally any more (usually) and this is also a warning for you in case
> you might once sent to other Debian lists on one hand people might
> respond annoyed if they get e-mails twice and on the other hand you
> might missing any response since you might not be subscribed to that
> specific list (posting policy is that everybody can post).  In this case
> I'd recommend reading the actual web archive.


Thanks for letting me know about this, I'll keep this in mind from now and
reply only to the list.


> On Fri, Feb 21, 2020 at 12:42:10PM +0530, Pranav Ballaney wrote:
>
> Hi Andreas,
> > Thanks for taking the time out to look into my work.
>
> You are perfectly welcome.  On one hand I see this as my duty as a
> (pontential) GSoC mentor.  On the other hand your level of contribution
> shows that you are really dedicated to the task and I'd love to support
> this.
>

Yes, it's my first time contributing to open source, so I'm really excited
about the opportunities this work provides, both of open source
contributions, as well as being part of something as prestigious as GSoC!


> > On Fri, Feb 21, 2020 at 1:54 AM Andreas Tille  wrote:
> >
> > > How are you running the test at your site?
> >
> > I run the tests inside a chroot I made using
> sbuild-debian-developer-setup
> > [1], and everything seems to be fine there as well.
>
> In principle an sbuild setup - provided that it also has something like
> a hook that is triggering the autopkgtest running in a chroot - should
> lead to the same results.
>

I just use the
autopkgtest .  -- schroot unstable-amd64-sbuild
command to run the tests. I'll try automating the process with pbuilder and
see if I can replicate the issue.
It will make things easier for me as well, for tests I write in the future.

> From the output, it seems that loki is trying to create a file called
> > loki.dat, but failing.
> > Just off the top of my head, could it be an issue with permissions? Does
> > the chroot allow programs to create files?
>
> Well, the script is copying files to that location - so file creation
> permissions should not be an issue, IMHO.  I also checked the code line
> in the source file that is throwing the error which looks not really
> suspicious to me.  But I only had a quick look.  Its pretty old code so
> we need to expect things that are not working as expected these days.


> > I will try building a pbuilder environment on my machine as well, and
> look
> > into this at the earliest.
>
> Just take your time.  In case you might be blocked it can help if you
> look into another package.  Liubov had announced to possibly have a look
> at the weekend and sometimes it helps to have an additional pair of
> eyeballs.  Sorry, for my "feature" to always stumble about issue. ;-)
>

Yes, I have started working on tests for fastlink, but it might take longer
this time because my mid-semester exams are starting, so I will be busy
with academics the next two weeks.


> > > Besides this issue you are perfectly on the right track and I repeat
> > > that I'm absolutely happy about your performance right from the
> > > beginning.
> >
> > Thank you so much!
>
> Again, you are really welcome
>
>Andreas.
>
> --
> http://fam-tille.de
>
> ᐧ


Re: Working tests seem to exist for epcr

2020-02-21 Thread Pranav Ballaney
Hi Andreas,

On Thu, Feb 20, 2020 at 8:40 PM Andreas Tille  wrote:

> Hi Pranav,
>
> On Thu, Feb 20, 2020 at 10:30:58PM +0530, Pranav Ballaney wrote:
> > Hi Andreas,
> > Thanks for looking into this and sending the guide about packages.
> > I saw the CI page for epcr, and I have a follow-up query about the same
> - I
> > had added autopkgtests to altree <
> https://salsa.debian.org/med-team/altree> the
> > other day, and I've now added to loki
> >  as well.
>
> Very good!  I've looked into altree meanwhile and will follow up with
> loki soon.
>
> > So, can CI pages be built
> > for these packages now? If yes, how?
>
> The pages are build for every new upload of a Debian package.  The upload
> can only be done by Debian Developers or Debian Maintainers.  Since you
> do not have this status you need to ask for sponsoring - well, your mail
> was kind of a request for sponsoring and so I did (at least for altree -
> loki will follow as I said).
>
Okay. I didn't know about this. So after I'm done writing tests for a
package, I'll just email this list (with the bug log in CC) and ask for
sponsoring?


> > Also, is it that only after the CI pages are built, should the respective
> > bugs (#909708 
> > and #909710 )
> be
> > marked closed?
>
> The bug is closed in the debian/changelog of the package.  Once the
> package is uploaded the bug will be closed.  This closes statement was
> actually missing in the epcr package - so that bug remained open.  I
> would like you to check my commits.  One is the changelog entry which
> closes the bug.  I would like you to do the following once you have
> written an autopkgtest:
>
> dch --team "Add autopkgtest (Closes: #bugnumber)"
>
> While you can perfectly edit debian/changelog with your favourite editor
> this simple line will do everything with the correct syntax for you.
>

Okay, I saw the commits and I'll do this as well for packages I work on in
future.


> I have also noted that you moved the example data from package
> altree-examples to the main altree package.  That's not a good idea
> since it somehow bloats a binary arch package with binary all data which
> are even not needed in normal installation of the package.  So I reverted
> this change and moved the data back to altree-examples.
>
> When doing so I realised that the test is failing due to compressed data
> files.  Debhelper has the habit to compress everything in
> /usr/share/doc/packagename that can sensibly be compressed and so its
> doing with some of the data files.  Uncompressing these in the test
> script helped here.
>

Thanks, I'll keep these things in mind from now.


> Finally I've added a debian/README.test which explains to the user how
> the package can be tested on the local machine.  It would be nice if
> you could drop such a file as well (you can find a template in our
> packaging template[1].
>

Okay, I'll add this file as well.


> > Or is it enough if the tests are working on my local
> > machine, and can I close them right away?
>
> As I explained, its not enough.  But your work was good enough to enable
> me to upload a package with autopkgtest after some polishing which is
> some routine work for me.  In other words:
>
>** Congratulations!  You have fixed your first bug and have your **
>** first package inside Debian!  That's great!   **
>
> > Thanks a lot for your help.
>
> You are perfectly welcome and I'm really happy about your work.  That
> was quite quick learning and a really good job for a beginner.  Keep
> on the good work
>

Thank you so much!


>
> Andreas.
>
>
> [1]
> https://salsa.debian.org/med-team/community/package_template/blob/master/debian/README.test
>
> > On Tue, Feb 18, 2020 at 8:35 PM Andreas Tille 
> wrote:
> >
> > > Hi Pranav,
> > >
> > > On Tue, Feb 18, 2020 at 05:07:44PM +0530, Pranav Ballaney wrote:
> > > > I am Pranav Ballaney, a Biology and Computer Science student from
> India,
> > > > and I have recently joined the Debian Med team, primarily writing
> > > > autopkgtests for various packages.
> > >
> > > Thanks for the short introduction here and the interest into the Debian
> > > Med GSoC / Outreachy project.
> >
> >
> >
> > > I tried running the tests present in this package, and they seem to
> > > > work well on my local machine. Can this bug be marked closed now?
> > >
> > > Ahhh, perfectly correct.  We simply forgot to close this bug in Debian
> > > changelog the string "Closes: #909706" (see here[1]).
> >
> >
> > > Since I'm new to Debian's development process, I'm not aware of the
> > > > procedure to mark a bug resolved. Does it just involve sending a
> mail to
> > > > 909706-cl...@bugs.debian.org?
> > >
> > > Exactly.  I'm doing this hereby (in CC).
> > >
> > > > I would really appreciate if someone could look into this for me. If
> any
> > > > more work is 

[RFS] python-depinfo

2020-02-21 Thread Nilesh Patra
Hi,
I have fixed python-depinfo which reports failed tests in debci[1] to pass
successfully.
It builds fine in a clean chroot with passing autopkgtests. I have also
updated to the next upstream release.
My changes have been pushed my changes here[2].
Needs review and sponsorship.

[1]:https://ci.debian.net/packages/p/python-depinfo/
[2]:https://salsa.debian.org/gi-boi-guest/python-depinfo

Thanks and regards
Nilesh


[RFS] predictnls

2020-02-21 Thread Nilesh Patra
Hi,
I have fixed predictnls which reports failed tests in debci[1].
It builds fine in a clean chroot with passing autopkgtests. I have pushed
my changes to the team repo, here[2].
Needs review and sponsorship.

[1]:https://ci.debian.net/packages/p/predictnls/
[2]:https://salsa.debian.org/med-team/predictnls

Thanks and regards
Nilesh


Re: Working tests seem to exist for epcr

2020-02-21 Thread Andreas Tille
Hi Pranav,

On Fri, Feb 21, 2020 at 07:37:45PM +0530, Pranav Ballaney wrote:
> >
> > The pages are build for every new upload of a Debian package.  The upload
> > can only be done by Debian Developers or Debian Maintainers.  Since you
> > do not have this status you need to ask for sponsoring - well, your mail
> > was kind of a request for sponsoring and so I did (at least for altree -
> > loki will follow as I said).
> >
> Okay. I didn't know about this. So after I'm done writing tests for a
> package, I'll just email this list (with the bug log in CC) and ask for
> sponsoring?

That's a correct summary, yes. :-) 
 
> > dch --team "Add autopkgtest (Closes: #bugnumber)"
> >
> > While you can perfectly edit debian/changelog with your favourite editor
> > this simple line will do everything with the correct syntax for you.
> 
> Okay, I saw the commits and I'll do this as well for packages I work on in
> future.
> ... 
> > Finally I've added a debian/README.test which explains to the user how
> > the package can be tested on the local machine.  It would be nice if
> > you could drop such a file as well (you can find a template in our
> > packaging template[1].
> 
> Okay, I'll add this file as well.

:-)
 
> > You are perfectly welcome and I'm really happy about your work.  That
> > was quite quick learning and a really good job for a beginner.  Keep
> > on the good work
> 
> Thank you so much!

You are welcome and thanks for your work

 Andreas.

-- 
http://fam-tille.de



Re: Failure in autopkgtest for loki (Was: Working tests seem to exist for epcr)

2020-02-21 Thread Andreas Tille
Hi Pranav,

On Fri, Feb 21, 2020 at 07:30:09PM +0530, Pranav Ballaney wrote:
> > You are perfectly welcome.  On one hand I see this as my duty as a
> > (pontential) GSoC mentor.  On the other hand your level of contribution
> > shows that you are really dedicated to the task and I'd love to support
> > this.
> 
> Yes, it's my first time contributing to open source, so I'm really excited
> about the opportunities this work provides, both of open source
> contributions, as well as being part of something as prestigious as GSoC!

I need to admit that your contributions had a high level and its hard to
believe for me that its your first time contributing to open source.
I'm always happy to see competent students in GSoC. 
 
> > In principle an sbuild setup - provided that it also has something like
> > a hook that is triggering the autopkgtest running in a chroot - should
> > lead to the same results.
> 
> I just use the
> autopkgtest .  -- schroot unstable-amd64-sbuild
> command to run the tests. I'll try automating the process with pbuilder and
> see if I can replicate the issue.
> It will make things easier for me as well, for tests I write in the future.

My personal experience with schroot is that I never managed to setup
sbuild to my satisfaction.  The configuration steps always seemed to be
quite complex.  Thus I sticked to pbuilder (which I by chance started
using first).
 
> > Just take your time.  In case you might be blocked it can help if you
> > look into another package.  Liubov had announced to possibly have a look
> > at the weekend and sometimes it helps to have an additional pair of
> > eyeballs.  Sorry, for my "feature" to always stumble about issue. ;-)
> >
> 
> Yes, I have started working on tests for fastlink, but it might take longer
> this time because my mid-semester exams are starting, so I will be busy
> with academics the next two weeks.

I can only repeat what I said to previous GSoC students:  Please always
give your study and your exams preference over GSoC work.  In case there
is urgent stuff for you to do just tell me and all is fine. 

Kind regards

   Andreas.

-- 
http://fam-tille.de