[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Re: Clarification of the project idea 'Enhancing Debian Blends web sentinel'

Hi Aditya,

On Tue, Apr 23, 2013 at 12:13:12PM +0530, Aditya Bhardwaj wrote:
> Hey..!!
> I am Aditya Bhardwaj, a 3rd undergraduate student at JIIT Noida, India
> enrolled in a Bachelor of Engineering course of Computer Sciences and
> Engineering.
> I have been going through the project ideas mentioned in Idea List of
> Debian for a while now and contacted Andreas Tille for the two
> project ideas that he is assigned a mentor for.
> Recently, I was interested in the project idea 'Enhancing Debian Blends web
> sentinel' and mailed him for the same. However, he seems to be occupied at
> the moment and it will take him time to reply to me.

Hmmm, didn't you got my reply to your first mail (the only one I did
received) and that I attached to this mail?  I responded even one day
before you did sended it (due to my freshly polished time travel device
or time-lag ... whatever explanation you prefer ;-)).  If you might have
sended another mail I'm sorry because this did not reached me for
whatever reason.  Sending to this mailing list is always safe ...

> I went through the documentation for blends at
> http://blends.alioth.debian.org/blends/ as mentioned by him a while back.
> And I must say that I am pleased with the idea of Blends as such.


> So, I would like some headers in order to begin with the project. Also, I
> would like to have more clarification regarding the objectives of the
> project.
> Firstly, regarding the 'refinement' of output by UDD, I would like to know
> how is the project looking to attain it? Is it by hacking UDD to produce
> better JSON output ( i.e. in case it is currently producing it ) or do we
> need to map existing an output format from UDD to JSON?

Currently the web sentinel pages are created as follows:

  1. UDD has a set of functions that are defined in


     You should look for
       ddtp_unique(text, text[])
       versions_archs_component (text)
       bibtex ()
     Seeking for the string "blends" in this code should lead you to the
     relevant pieces (which might be a bit "hidden" amongst unrelated UDD things)

  2. The script

     calls these functions to obtain the needed information and puts
     these into classes of a very weak (historically hacked) design for
     the purpose it should serve now.

  3. The assembled data will be rendered into html by using Genshi
     templates available here:


My first approach was to write another set of templates to create JSON
output from the gathered data but this failed because Genshi needs XML.

> In case its possible, I would like some clarification as to how its
> currently done.

That's the current state.  I have read that latest PostgreSQL (9.2) has
new JSON features and thus it might come to mind to try direct JSON
output directly from PostgreSQL (I never checked JSON with PG 9.1).  If
this is feasible we could probably save a lot of overhead.

> Secondly, I would like know how the Blends web sentinel must be enhanced.

Currently there were the following feature requests:

  1. Reusing Blends data / different rendering:  The request came from
     NeuroDebian people who would like to render their pages (for instance
     like this


     from the same source (UDD) as the Blends sentinel.  Actually the
     NeuroDebian pages have something in common with the web sentinel and
     NeuroDebian in itself in principle qualifies as Blend - but they do
     not (yet) use the common techniques (because they have a certain
     opinion on how should things look like).  The goal is to make the
     web sentinel good enough / fitting for NeuroDebian to settle with
     one common technique. (Yaroslav, do you want to comment on this?)

  2. Design:  Hmm, whatever this might mean - I'm no web designer.  The
     people who asket for "better" design did not specify what they
     actually mean by "better".  What I personally do not like is the
     way you can get extra data about versions and architectures or the
     screenshots when using the mouse over feature.  This is a bit weak -
     as I said I'm no web designer.

  3. User Response: There were users who wanted to give some response about
     the packages.  I'm personally not very convinced about this because
     I have no idea how to "react" on this kind of response and whether we
     just open the door for another way of spamming.  But this needs to be

> >From what I can gather, it will be implemented via the functionality
> provided by the first part of the idea ( which shall give it the necessary
> JSON to work on ). Also, what must be the underlying architecture of such a
> portal? Will it have Django or another python based 'engine'? Or another
> architecture altogether?

The current technique are static HTML pages.  I personally do not have
any experience with Python based frameworks (except with Zope from about
ten years ago but this is history).  So in principle Django seems to be
fine.  I'm personally a bit confused by all those "Python frameworks of
the year" which are hyped for some time and than turn out as "just
another framework amongst several others.  I hope for other educated
input to help in the decision for some framework that fits.  The only
hard fact is that it needs to be possible to be installed on machines
maintained by DSA[1]

> I was also hoping if someone could clarify me the availability status of
> the same.

Feel free to keep on asking if you need further information.  I'd be
happy to mentor you in the GSoC project and if there was some loss of
e-mail I'd be sorry about this and hope this is fixed by this answer.

Kind regards


[1] http://wiki.debian.org/Teams/DSA

Date: Mon, 8 Apr 2013 23:15:21 +0200
From: Andreas Tille <andreas@an3as.eu>
To: Aditya Bhardwaj <aditya2204@gmail.com>
Subject: Re: [GSoC 2013] Additional headers for the projects 'Redesign
 metapackage creation for Debian Blends' and 'Enhancing Debian Blends web
Message-ID: <20130408211521.GB30524@an3as.eu>
References: <CANr_U0L57342bhWxo0dD8As6V1CZsQAvdHUEVvGgH992YcmQLQ@mail.gmail.com>
MIME-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
In-Reply-To: <CANr_U0L57342bhWxo0dD8As6V1CZsQAvdHUEVvGgH992YcmQLQ@mail.gmail.com>
User-Agent: Mutt/1.5.20 (2009-06-14)
Status: RO

Hi Aditya,

thanks for your interest in the GSoC projects.

On Tue, Apr 09, 2013 at 02:03:50AM +0530, Aditya Bhardwaj wrote:
> I am Aditya Bhardwaj, a 3rd undergraduate student at JIIT Noida, India
> enrolled in a Bachelor of Engineering course of Computer Sciences and
> Engineering.
> I have been going through the project lists on the Debian site:
> http://wiki.debian.org/SummerOfCode2013/Projects and came across these
> project ideas, 'Redesign metapackage creation for Debian Blends' and
> 'Enhancing Debian Blends web sentinel'.
> Now, I was basically looking for Python based projects and they seem
> interesting. So, I was hoping for you to provide me with additional headers
> for the same so that I could begin some preliminary research on it. Or even
> for setting up an environment so that I could start getting
> acquainted/started with the same.

I guess the best starting point is to read


to become comfortable with the Blends idea.

> As far as my coding skills are concerned.. I have worked using Python, C,
> C++, Java.. Know some bash scripting and have also done some Android
> development.. Done GUI through GTK and Qt (via Python and C++).. Used
> 'Quickly' as well..!!
> I am also a moderately-old Linux user and a FOSS enthusiast. I have also
> contributed a patch to Mozilla, so I am aware of the basic nuances of
> open-source software development.

Sounds good - I'd be happy to work together with you in case the project
will be accepted.
Just let me know if you need more detailed information.

Kind regards



Reply to: