[OSM-dev] Google Summer of Code

[Gora Mohanty]

(Sorry for the long, unedited follow-up, but I wanted to keep context.)

Couple of basic questions on OSM GSoc projects:
(a) Wi.l there be a publicly available repository for the code?
(b) What technology does the OSM website use? RoR? Is it possible
    to get a local installation of the website, if not of all the
    data? Sorry, if things like this have been addressed in the
    GSoc proposal. I don't know where to look offhand,
(c) Arindam, where are you based? If you are somewhere near Delhi, it
    would be great if we could meet face-to-face.

On Wed, 23 Apr 2008 11:50:32 -0700 (PDT)
Mikel Maron <mikel_maron at yahoo.com> wrote:

> 
> > From: Tom Hughes <tom at compton.nu>
> > 
> > In message <8990327d0804230928p7739f912o568fdf24020896ad at mail.gmail.com>
> >          "Arindam Ghosh" <makghosh at gmail.com> wrote:
> >
> > > On Tue, Apr 22, 2008 at 3:22 PM, Gora Mohanty <gora at sarai.net> wrote:
> > > >  I work on internationalisation, and localisation issues, especially
> > > >  as these pertain to Indian languages, and would be glad to help out
> > > >  on this project. I hope that there is no need for me to become an
> > > >  official mentor, as I cannot always guarantee having time available
> > > >  when needed.
> > > 
> > > That's great...:) Can you please provide some pointers regarding
> > > implementation of lingatures and vowel positioning of Indic scripts. I
> > > would like to study how much of these would be required pertaining to
> > > project.
> > 
> > I wouldn't have thought you should have to worry about that much
> > really - if you've got valid UTF and you use an appropriate text
> > rendering library then that sort of stuff should sort itself out.
> 
> Rendering Indic characters into glyphs is not so straight forward --
> the fonts require a lookup table or state machine to account for all the
> variations in character combination.

That is true, but in most cases it should be possible to hand off this
work to lower-level rendering engines. And, while I love the Indic focus,
we should have a broader agenda of complete Unicode compliance.

> For the web site, yes I think we can rely on the browser for this, with testing
> of course.

Internationalising the website might range from quite easy to difficult.
Ruby kind of sucks at handling UTF-8. Django, and Java are much easier.

> For other pieces of the ecosystem, it will need more inquiry, if there's time
> this summer. I am aware that IndicTrans has been working on an open source
> solution, but not sure how much work it will be to integrate.

Could you define what the pieces of the ecosystem are?  I have been playing
around with GeoDjango (the GIS branch of Django), and while it is still in
development, the mapping and GIS aspects rock. Also, from what I see,
localisation of things like tags should be trivial with GeoDjango, though I
really should try it out before spouting off.

Regards,
Gora