Re: GSoC 2015 - should there be a Debian Med / Blend proposal separate from Debian?
Hi Steffen,
On Tue, Oct 07, 2014 at 05:13:06PM +0200, "Steffen Möller" wrote:
> Hello,
>
> The projects of Debian for the Google Summer of Code
> http://www.google-melange.com/gsoc/homepage/google/gsoc2015
> have their focus on the Debian infrastructure. It was quite some earning
> experience for me, but eventually I have grasped that this was a good thing.
> However, my personal ambition for Debian comes for a good part from the
> practical side of it all - it is my Bioinformatics desktop and/or server - and
> I want that helped by a GSoC, too.
>
> Now Debian is good (best?) with sharing libraries between otherwise
> independent software packages, but we are yet doing little to help
> with the inter-operability of those packages. And beyond our Wiki there
> is nothing that informs others about how to address our daily routine.
> This GSoC could help Debian (as a whole, not just "Med") to strengthen
> its ties with upstream and foster developments to help with the
> interaction of tools and the promotion of such workflows.
>
> While I see naturally quite some opportunities in Bioinformatics, I
> am highly interested to see our other blends address such workflows,
> too. So this could be a Debian Blend Project. Here some ideas:
>
> * Get Gentle (molecular cloning) dissected into several interoperating
> small command line tools and combine it with a workflow environment
>
> The reason for that is that molecular cloning is an iterative process
> and Gentle (like all other tools for molecular cloning I looked at,
> correct me if I am wrong) are dysfunctional with a representation of
> the steps taken to yield a particular amplicon. The workflow is the
> recipe to pass to the technical assistant. The workflow engine could
> be Taverna as a start, which is offered as a Debian package by upstream
> and ... well ... some day my package for it will work, there may be
> others.
>
> * Functional Tutorials
>
> We started to look at Youtube to learn about how to use any particular
> software. This hurts. Just a bit. I would very much like to see something
> in the lines of how Docker or VIM introduces itself as a template for
> biological sequence analysis.
>
> * Unit Tests
>
> Andreas has already been much at it. There should be more. And more.
> This GSoC could help. I reckon that we could also come up with new ideas
> about what such tests could work like, for instance when there is
> no perfect answer to a problem, as in the assembly of complete genomes
> from short sequences.
>
> * FPGA for application acceleration
>
> It was amazing to see how quickly the BitCoin folks jumped from GPUs
> to FPGA to ASICs. There is quite a number of affordable FPGA boards
> now available that would be good to have closer to our distribution
> to prepare for application acceleration. This fancy Open Source Laptop
> https://www.crowdsupply.com/kosagi/novena-open-laptop
> even has an FPGA within - albeit quite a small one. I would very
> much like to see something happening to help sharing Open Source
> hardware descriptions between such devices.
>
> ...
>
> There is more. Much more. But how is your feeling about it all? Is it a
> good idea to have this organised in parallel to what the core of Debian
> is doing? Are such worthwhile projects in the first place? Are there
> other GSoC organisations that are likely to care for them?
I agree that there are several interesting points but I for myself
regard the pure time spent into the mentoring itself as long enough. I
do not want to spent it in the additional bureaucratic stuff. So I will
not stop anybody with an honest attempt to do it but we probably need to
be prepared to answer the question why a Debian subgroup wants to
register as extra organisation. If I where in charge from the Google
side of GSoC I would ask this question.
Moreover your description of projects above IMHO fits to some extend
to the general policy Debian has for GSoC.
BTW I would have expected something for Outreach Program for Women. I
personally have put all my forces to make sure that all our packages are
migrating to testing (Steffen, you remember that I asked for somebody
else filing the needed ROM bug for mgltools-sff, right (hint, hint)) and
so I was not really up for finding a proper project description. Any
volunteer for this?
Kind regards
Andreas.
--
http://fam-tille.de
Reply to: