Re: [boost] [SoC] Summer of Code Project Ideas

Subject: Re: [boost] [SoC] Summer of Code Project Ideas
From: Mateusz Loskot (mateusz_at_[hidden])
Date: 2010-03-09 14:44:35

Andrew Sutton wrote:
>> Within Boost.Geometry, we have discussed participation in the GSoC a
>> bit. However, we've decided to not to propose or mentor any projects
>> due to amount of work we need to get done ASAP to include the library in
>> one of the upcoming Boost releases.
>>
>> I have mentored one GSoC project myself (though, not for Boost)
>> and I would not be able to commit to mentoring while working on the
>> library preparation at the same time.
>>
>> However, I personally thing, that if someone else would be interested in
>> mentoring a student working on some project for Boost.Geometry, that
>> would be great.
>
> 
> That's too bad. I was actually thinking that this library is ideal for SoC
> projects. 

Yes, I understand it.

> A good student might be able to work on extensions to your
> library that don't affect primary submission, but can be integrated after
> eventual acceptance.

The problem is not with affecting the Boost.Geometry submission, but the
problem is with amount of work required to make the submission happen
and our time. As we've discussed internally in team, we need to focus
on the after-review tasks first.

> I'm not sure that anybody else would be a better mentor. It's your library,
> so you have have all the knowledge and expertise :)

I assumed, perhaps a bit naively :-), that there may be some developers
out there who are already getting well familiar with Boost.Geometry
internals and perhaps would be willing to mentor a student.

Certainly, I would do my best to provide some support, but I'm not
able to offer myself as "full time" mentor before we release the library.

>>> == Infrastructure Projects==
>>
> What could that be?
>> Compilation farm? Developing regression resting?
>> May be something like integrating Boost regression tests
>> with existing systems like Buildbot, Hudson, etc.
>> This could increase potential of attaching new test environments.
>>
> 
> Vladimir Prus has some suggestions on the Boost wiki (
> https://svn.boost.org/trac/boost/wiki/SoC2010#Boost.Build).

I'll post comments to it separately.

>> It could also include projects for integrating/collaboration between
>> Boost libraries. For instance, number of ideas have been discussed
>> about integrating Boost Geometry and Boost Polygon libraries in terms of
>> using types of one with algorithms of the other one, etc.
>>
>> Another interesting proposals could be projects integrating GIL
>> (rasters) with Geometry (vectors). For example, implementation of
>> algorithms for rasterisation of vectors or vectorization of rasters
>> seem to be ideal candidates for extensions.
>>
> 
> I think those are great ideas, but they may be a little broad of scope for
> student projects.

Yes, but I think I gave a broad overview.
A student could work on small part, a specific function.
For example, to polygonize Boost.GIL raster by creating Boost.Geometry
vector for all connected regions of pixels in the raster sharing a
common value, It should be feasible in the given time frame.

> I think those proposals might also tend toward the
> experimental, and while interesting may not have the publishable payoff that
> some students are looking for. Still... very good ideas.

Good point.

Best regards,
-- 
Mateusz Loskot, http://mateusz.loskot.net
Charter Member of OSGeo, http://osgeo.org