headers
Lucas Nussbaum | 1 Mar 2008 01:55

Re: Google Summer of Code 2008

On 29/02/08 at 19:55 +0000, Steve McIntyre wrote:
> Lucas wrote:
> >On 28/02/08 at 01:09 -0800, Steve Langasek wrote:
> >> 
> >> Yes, subjective to the point of absurdity.  If failure is defined in terms
> >> of *your* expectations, I don't see how we can even have a meaningful
> >> dialogue about it.
> >
> >Note that my main point in the thread is "we should use GSOC to get
> >fresh blood in Debian, not to fund existing contributors". The point
> >about "Debian GSOC projects have been unsuccessful in the past" is
> >totally secondary.
> >
> >I am under the impression that results from last years' GSOC projects
> >weren't up to par with what could have reasonably been expected from
> >them, based on the skills of the students and the time they were
> >supposed to spend on the projects. Maybe I'm wrong, but it will be
> >difficult for you to convince me of that, since we lack data :-)
> 
> But that's not going to stop you making accusations of previous GSoC
> students and mentors misleading Google about how time was spent,
> though. That's *nice* to see.

I think something. You think something else. There's no data to back
either claim, so we just have to live with it.

Note that the whole "did last year projects were successful?" issue is
secondary. Even if all of last years projects produced fabulous results
that totally changed the way Debian is developed, I'm still not sure if
we should use GSOC to pay current Debian contributors, instead of using
(Continue reading)

Permalink | Reply | 3 comments |
headers
Shaun Jackman | 1 Mar 2008 02:19
Picon
Gravatar

Re: binary vs "real debian" packages

On Fri, Feb 29, 2008 at 3:05 PM, William Francis <wfrancis <at> gmail.com> wrote:
...
>  Further, I understand the concept of an upstream provider and
>  understand that I don't have one in this case, unless I sort of fake
>  it somehow. Is that wise or is there a well understood method of
>  having an .orig file and then doing stuff to make your .dsc, .diff and
>  .changes files? What would the contents of an orig file like that look
>  like in my case where it's not a source package?

Hi William,

A quick answer to one of your questions...
The orig file would contain all the files not in the debian/
directory, and the diff file would contain all the files in the
debian/ directory.

Cheers,
Shaun
Permalink | Reply | 1 comment |
headers
Cameron Dale | 1 Mar 2008 03:08
Picon

Re: Google Summer of Code 2008

On 2/29/08, Lucas Nussbaum <lucas <at> lucas-nussbaum.net> wrote:
> Do you have any proof that GSOC students worked 35-40 hours a week on
>  their GSOC projects? You probably don't. So again, no real data to back
>  either claim. We have different opinions, and have to live with it.

I don't think there's anything in the GSoC that requires 35-40 hours
per week. From some of the postings on the students list last year it
seems like 20 hours per week is more common. I think Google leaves
this up to the mentoring organization and students to work out for
themselves, so if we require 35-40 hours per week, we should obtain
assurances from the students during the application process that they
have that time to commit.

>  Frankly, if I were in the position of a GSOC student, I would probably
>  find it very hard to work 35-40 hours per week on my project, while I
>  could squash some items off my TODO list. Maybe the whole problem is
>  that I'm less disciplined than our students ;)

I was a GSoC student for Debian last year. I estimate I put in close
to 35 hours per week, but it may have been closer to 30. This year I
don't plan on applying as I'm finishing my thesis, though by your
suggestion I would not be accepted anyway as I am at the DAM stage of
NM.

I also maintain several packages, and was in the NM queue when I
applied to GSoC last year. I consider packaging to be a different
style of contribution than my GSoC project, as all my packages were
just packaging, while my project is my own code (and now a 'native'
package). I certainly did work on my other packages during the summer,
but just like this work doesn't interfere with school or full-time
(Continue reading)

Permalink | Reply | 0 comments |
headers
Ben Finney | 1 Mar 2008 04:27
Picon

Re: binary vs "real debian" packages

"Shaun Jackman" <sjackman <at> gmail.com> writes:

> The orig file would contain all the files not in the debian/
> directory, and the diff file would contain all the files in the
> debian/ directory.

More accurately, the 'foo-1.2.3.orig.tar.gz' file would contain the
"upstream from the perspective of Debian" source; that is, the working
tree of all files that someone would want if they were taking the
software and uninterested in making a Debian package.

The 'foo_1.2.3-1.diff.gz' would contain whatever changes are necessary
to turn that working tree into something buildable to a Debian binary
package.

The 'foo_1.2.3-1.dsc' would contain necessary metadata about the
combination of 'foo-1.2.3.tar.gz' and 'foo_1.2.3-1.diff.gz'.

To answer the OP's "for some reason" implied question: the combination
of these three files is a "Debian source package".

-- 
 \        “To punish me for my contempt of authority, Fate has made me |
  `\                an authority myself.” —Albert Einstein, 1930-09-18 |
_o__)                                                                  |
Ben Finney

--

-- 
To UNSUBSCRIBE, email to debian-devel-REQUEST <at> lists.debian.org
with a subject of "unsubscribe". Trouble? Contact listmaster <at> lists.debian.org
(Continue reading)

Permalink | Reply | 0 comments |
headers
Charles Plessy | 1 Mar 2008 04:36
Favicon

Buildd backlog and testing transition.

Le Fri, Feb 29, 2008 at 10:40:57AM +0100, Marc 'HE' Brockschmidt a écrit :
> 
> Due to kernel problems, the mips* buildds haven't been very reliable in
> the past few weeks, creating a loooong backlog of packages that need to
> be built. As there seems to be a workaround for the kernel bug, this
> should start getting better from the weekend on. As a maintainer: Just
> wait.

Dear Marc,

it is good news to read that there is a solution being found. However, I
am a bit confused because previous messages were suggesting that the
problem was disk speed, not downtime.

In the meantime, could the release team do us the favor of letting the
packages migrate? I just got a message from a user who wants to use one
of the blocked packages (emboss), and I am just so ashamed to answer him
that he has to use unstable just because it is not built on a platform
where nobody is using it.

Please. Our work is left aside for no benefit, and if I understand
correctly, transiently allowing the migration is just a matter of
changing a configuration file two times.

Have a nice day,

--

-- 
Charles Plessy
http://charles.plessy.org
Wakō, Saitama, Japan
(Continue reading)

Permalink | Reply | 29 comments |
headers
Charles Plessy | 1 Mar 2008 05:27
Favicon

Re: binary vs "real debian" packages

Le Fri, Feb 29, 2008 at 03:05:58PM -0800, William Francis a écrit :
> 
> my contents are not source (configure, make, etc), rather I'm more
> interested in the preinst/postinst scripts, the Depends part of the
> control file, a few config files and placing a few scripts on the
> filesystem that require no compiling.

> Further, I understand the concept of an upstream provider and
> understand that I don't have one in this case

Dear William,

If you do not plan to release the packaged files separately, you may be
interested in the "native" format of source packages, in which the
debian directory and the upstream files are contained in the same tar
archive. The unpacked source package looks the same be it native or not,
so general tutorials are good readings anyway. You can refer to the new
maintainer guide for instance:

http://www.debian.org/doc/maint-guide/

If all you have to do is move files, you can shortcut a lot of the
complexity by using the common debian build system (CDBS):

http://perso.duckcorp.org/duck/cdbs-doc/cdbs-doc.xhtml

If you have more questions, they may be more relevant on the
debian-mentors <at> lists.debian.org mailign list:

http://lists.debian.org/debian-mentors/
(Continue reading)

Permalink | Reply | 0 comments |
headers
Steve Langasek | 1 Mar 2008 08:29
Picon
Favicon

Re: Google Summer of Code 2008

On Sat, Mar 01, 2008 at 01:55:49AM +0100, Lucas Nussbaum wrote:

> Note that the whole "did last year projects were successful?" issue is
> secondary. Even if all of last years projects produced fabulous results
> that totally changed the way Debian is developed, I'm still not sure if
> we should use GSOC to pay current Debian contributors, instead of using
> it to bring in new contributors.

So you think it's better to focus on people who aren't sufficiently
motivated to get involved in Debian without being paid to do so?

> > >> I'm not saying that students that were DD did nothing of their time
> > >> during GSoc, but most of them produced results that were a bit
> > >> disappointing given what people could have expected from them, mainly
> > >> because they used their GSOC time to work on other Debian tasks.

> > Do you have any proof at all for that accusation? If so, please share
> > it. Otherwise, I think that people deserve apologies from you right
> > now.

> Do you have any proof that GSOC students worked 35-40 hours a week on
> their GSOC projects? You probably don't. So again, no real data to back
> either claim. We have different opinions, and have to live with it.

Where does this "35-40 hours" figure come from?

  10. How much time is required to participate as a student in Google Summer
  of Code?

  The amount of time you will need depends on both the scope of your project
(Continue reading)

Permalink | Reply | 1 comment |
headers
Andreas Tille | 1 Mar 2008 10:05
Picon
Favicon

Re: How important is "Architecture: any" (Was: Downloads things, ...) (fwd)

Ups, sorry, wrong list debian-devel.  Should have gone to debian-edu ...

-- 
http://fam-tille.de

---------- Forwarded message ----------
Date: Sat, 1 Mar 2008 00:42:02 +0100 (CET)
From: Andreas Tille <tillea <at> rki.de>
To: Debian Developers <debian-devel <at> lists.debian.org>
Subject: Re: How important is "Architecture: any" (Was: Downloads things, ...)
Resent-Date: Fri, 29 Feb 2008 23:43:13 +0000 (UTC)
Resent-From: debian-devel <at> lists.debian.org

On Sat, 1 Mar 2008, Petter Reinholdtsen wrote:

> I guess you might be right, now.  Earlier, we used depends to pull in
> packages using the meta packages during installation.  Now we use
> tasksel tasks, which are more forgiving about missing packages.  We
> did have a problem with the debian-edu package failing to propagate
> into testing because it depended on arch-any packages that was missing
> on some architecture.  Now that we are using recommends, and
> recommends might even be installed by default, I guess the need is not
> so high.

Hmmm, once you say so I remember (non-RC!) bugs for not available
Recommended packages.  So it will not block anything but I'm not sure
about the policy here.  If we are just asking for those bug reports
(even if non RC) the plan is quite suboptimal.

Kind regards
(Continue reading)

Permalink | Reply | 0 comments |
headers
Lucas Nussbaum | 1 Mar 2008 11:40

Re: Google Summer of Code 2008

On 29/02/08 at 23:29 -0800, Steve Langasek wrote:
> On Sat, Mar 01, 2008 at 01:55:49AM +0100, Lucas Nussbaum wrote:
> 
> > Note that the whole "did last year projects were successful?" issue is
> > secondary. Even if all of last years projects produced fabulous results
> > that totally changed the way Debian is developed, I'm still not sure if
> > we should use GSOC to pay current Debian contributors, instead of using
> > it to bring in new contributors.
> 
> So you think it's better to focus on people who aren't sufficiently
> motivated to get involved in Debian without being paid to do so?

Motivation isn't the only thing necessary to get involved in a free
software project. It seems to me that often, people don't participate
because they don't find a good answer to the "How can I help?" question.
Debian isn't really good at answering this question, other projects
(GNOME, Ubuntu come to mind) do much better than us.

> > > >> I'm not saying that students that were DD did nothing of their time
> > > >> during GSoc, but most of them produced results that were a bit
> > > >> disappointing given what people could have expected from them, mainly
> > > >> because they used their GSOC time to work on other Debian tasks.
> 
> > > Do you have any proof at all for that accusation? If so, please share
> > > it. Otherwise, I think that people deserve apologies from you right
> > > now.
> 
> > Do you have any proof that GSOC students worked 35-40 hours a week on
> > their GSOC projects? You probably don't. So again, no real data to back
> > either claim. We have different opinions, and have to live with it.
(Continue reading)

Permalink | Reply | 0 comments |
headers
Florian Lohoff | 1 Mar 2008 10:45

Re: Buildd backlog and testing transition.

On Sat, Mar 01, 2008 at 12:36:40PM +0900, Charles Plessy wrote:
> it is good news to read that there is a solution being found. However, I
> am a bit confused because previous messages were suggesting that the
> problem was disk speed, not downtime.

Downtime caused by ghc6 build causing multiple kernel crashes on mips and mipsel.
Newer kernel which might fix the issue are not available due to a kernel
bug (#466977) since 2.6.18 ... (Running 2.6.17).

To have the buildds catch up faster i as a buildd host was asked to
provide a faster disk subsystem which i did...

So you might devote your time to 

a) Find the cause of the build crash
b) Hunt down the kernel bug in 2.6.24
c) Poke at the buildd admin to move the buildd to the new disk subsys

All those might actually solve the problem and not work around it ..

Thanks for you time ...

Flo
--

-- 
Florian Lohoff                  flo <at> rfc822.org             +49-171-2280134
	Those who would give up a little freedom to get a little 
          security shall soon have neither - Benjamin Franklin
Permalink | Reply | 1 comment |

Gmane