[gentoo-project] New research: Gentoo Portage Package Dependencies

[gentoo-project] New research: Gentoo Portage Package Dependencies

[Pavlos Ratis]

Hello,

Recently I found that there's a new research on Gentoo, specifically
for Portage and package dependencies[1].

-- Authors --
Remco Bloemen, Chintan Amrit, Stefan Kuhlmann, Gonzalo
Ordóñez–Matamoros  -- University of Twente

-- Abstract --
  "Open source distributions such as Gentoo need to accurately track
dependency relations between software packages in order to install
working systems. To do this, Gentoo has a carefully authored database
containing those relations. In this paper, we extract the Gentoo
package dependency graph and its changes over time. The final
dependency graph spans 15 thousand open source projects and 80
thousand dependency relations. Furthermore, the development of this
graph is tracked over time from the beginning of the Gentoo project in
2000 to the first quarter of 2012, with monthly resolution. We perform
a cluster analyses of the package dependency graph that reveal
meaningful relations among packages, and in a separate paper we
analyse changes in the dependencies over time to get insights in the
innovation dynamics of open source software. The resulting dataset
provides many opportunities for both research on Open Source as well
as practice."

 The research paper is going to be presented in the 11th Working
Conference on Mining Software Repositories[2].

[1] http://www.researchgate.net/publication/261100411_Gentoo_Portage_Package_Dependencies

[2] http://2014.msrconf.org/program.php

Regards,
Pavlos

Re: [gentoo-project] New research: Gentoo Portage Package Dependencies

[Rich Freeman]

On Mon, May 19, 2014 at 6:24 AM, Pavlos Ratis <dastergon@gentoo.org> wrote:
> Recently I found that there's a new research on Gentoo, specifically
> for Portage and package dependencies[1].

Thanks for pointing this out!  Anybody know anything about their first
reference, also from 2014?  As far as I can tell it hasn't been
published yet but I figured I'd ask here before I go bugging the
authors.

Interesting that they picked Gentoo.  It might just be from
familiarity, but a reason to use Gentoo is that the dependency info
for a package is separated from the actual distfiles/etc.  Many other
package formats combine these, though perhaps this data is cached
somewhere in a format that could be used historically.  If they had
multiple GB of ebuild files I'd hate to think about what that would
look like as .debs.

Rich

Re: [gentoo-project] New research: Gentoo Portage Package Dependencies

[Patrick Lauer]

On 05/19/2014 06:24 PM, Pavlos Ratis wrote:
> Hello,
> 
> Recently I found that there's a new research on Gentoo, specifically
> for Portage and package dependencies[1].
> 

[snip]

So that's quite ... hmm, I don't know.

There's an interesting lesson to be found: If you want to do such
research, get in contact with the developers. I was quite amused that
they spent literally weeks fetching things from anoncvs.g.o instead of
being lazy, or using some weird hackarounds instead of reading the
metadata cache directly.

For me such things are obvious, for them it's magic - but apparently
they never bothered to ask.

The result seems to be that libraries only get used when they exist and
not before, and/or I'm too critical of conclusions. Might be nice to
make some more graphs showing how things evolve over time.

Re: [gentoo-project] New research: Gentoo Portage Package Dependencies

[Vikraman Choudhury]

[-- Attachment #1: Type: text/plain, Size: 1095 bytes --]


Rich Freeman <rich0@gentoo.org> writes:

> On Mon, May 19, 2014 at 6:24 AM, Pavlos Ratis <dastergon@gentoo.org> wrote:
>> Recently I found that there's a new research on Gentoo, specifically
>> for Portage and package dependencies[1].
>
> Thanks for pointing this out!  Anybody know anything about their first
> reference, also from 2014?  As far as I can tell it hasn't been
> published yet but I figured I'd ask here before I go bugging the
> authors.
>

I found this after googling:

http://essay.utwente.nl/61920/1/Remco_Bloemen_-_Innovation_Dynamics_in_Open_Source_Software.pdf

It is a more interesting read compared to the paper.

> Interesting that they picked Gentoo.  It might just be from
> familiarity, but a reason to use Gentoo is that the dependency info
> for a package is separated from the actual distfiles/etc.  Many other
> package formats combine these, though perhaps this data is cached
> somewhere in a format that could be used historically.  If they had
> multiple GB of ebuild files I'd hate to think about what that would
> look like as .debs.
>
> Rich

--
Vikraman

[-- Attachment #2: Type: application/pgp-signature, Size: 359 bytes --]