[OSGeo-Discuss] Small projects

[Cameron Shorter]

Chris Holmes wrote:
> 
> 
> Frank Warmerdam wrote:
> 
>> Ari Jolma wrote:
>>
>>> Is there any idea in promoting projects, which develop small,
>>> well-defined, interoperable tools? Is my impression correct that the
>>> process that's been thought of for projects to join the foundation is a
>>> bit heavy and perhaps not suitable for small projects? I believe
>>> promoting interoperability is one of the foundations aims.
>>
>>
>>
>> Ari,
>>
>> I have been asking myself this question too.  The "Project" 
>> infrastructure,
>> requiring a PSC, an incubation period, committer agreements, web 
>> sub-domain
>> and so forth seems pretty heavy.  So, I am doubtful that it would be
>> appropriate to handle more modest sized project that way.   I have been
>> asking myself a similar question about things like libtiff, libgeotiff,
>> PROJ.4, MITAB, AVCE00 and Shapelib.  My conclusion so far is that they
>> might be better handled as a sub-project of an existing major project
>> (such as GDAL for the above file translators).   I am assuming that a
>> sub-project would not need a PSC, but would be subject to the PSC of
>> the project it is part of and mostly would just act as a part of that
>> project as far as the foundation is concerned.
> 
> I've been thinking about this approach as well.  Apache definitely works 
> that way, with db and jakarta.  And I think would work well for some of 
> the sub-projects of the more established communities.  It also might 
> make sense to have a 'front end' project, that groups together mapbender 
> and mapbuilder, and other lighterweight web clients could also go in to 
> that group.  Not that they wouldn't each have their own PSC, but such a 
> grouping might be nice.  Something like quickWMS could live in that 
> group without a PSC, for example.

What I like about the Apache style is that the look for projects which 
will grow and will be supported when the author or sponsoring company 
looses interest.

So for all projects, I'd be looking to find a community who will 
continue to maintain the project when the author(s) move on.  In the 
case of a library, the projects which use the library are likely to take 
over the maintainance of that project.  In this case the library could 
be bundled with the project which uses it.

However, I don't think it makes sence to bundle similar projects unless 
they share each other's code.  Eg, quickWMS shouldn't be bundled with 
Mapbuilder just because they do the same sort of thing.  If the 
author(s) of quickWMS looses interest, Mapbuilder developers are 
probably not going to take over the development of quickWMS.  (They will 
be more interested in the Mapbuilder code).

-- 
Cameron Shorter
http://cameron.shorter.net