Geospatial Code Sprinting

Jason Birch writes:

P.S. OK, one thing annoys me more than non-contributing users: businesses that use open source software and give nothing back. If you're leveraging open source software for competitive advantage, spin some of that back to the development community in code or $$$. Something as simple as contributing space, cola, and pizza for a code sprint can make a huge difference.

The open source GIS community, with a few exceptions, hasn't done code sprints. I used to advocate for MapServer code sprints at past meetings, but there was little interest. OpenLayers is one project that could have successful community sprints. There are a lot of energetic, can-do programmers, a solid testing framework, and all you need is a text editor and a modern web browser. My understanding is that MetaCarta already has in-house sprints of a sort, and these could someday be expanded to include community developers.

Comments

Re: Geospatial Code Sprinting

Author: Paul Ramsey

One thing (but not the only thing) we are going to add to FOSS4G is a day at the end for code sprints and project team meetings. Self organized, self directed, we'll provide a conference room and internet connectivity, the projects provide the rest. Watching the Geotools team spend last years conference sprinting on the exhibition floor made me say "there has to be some way these guys can do this without missing the whole conference".

Re: Geospatial Code Sprinting

Author: Christopher Schmidt

The biggest problem with code sprints for OpenLayers is actually getting everyone in one place. Even within MetaCarta, it's difficult, which is why they probably seem like sprints in the first place: Basically, any time that Erik is in town, the three of us hack on OpenLayers. This leads to a major development push about every 3 months, since that's how often he flies through. There are no other major developers of OpenLayers within 4 hours, though, which means that offering the space for a sprint is less useful. I'm sure that if we had anyone else in town other than Schuyler and I who were working on code, we'd be happy to have them around, and typically the weeks that we have developer time, I spend a lot of time on IRC trying to make sure we're doing the right thing with all the patches we have outstanding. It's not a perfect solution, but geographic distance provides a lot of difficulties -- communication and logistics chief among them -- to the extent that I'm not convinced that a company can actually offer space for a code sprint effectively without being near a conference location to begin with.

Re: Geospatial Code Sprinting

Author: Jason Birch

Hmm. Guess I should have added in airfare and accommodations :) I see your point. Is there anything else that businesses can do that is easy for them and typically difficult or costly for the community to do on its own? Jason

Re: Geospatial Code Sprinting

Author: Andy

When I worked for AT&T which later became Comcast and now I hear it became Cingualr or Cingular became it, whatever I can't keep track of it anymore. Anyway at AT&T we used a lot of open source and we would frequently pick functionality that we thought we would like or that we thought the community in general would like and we would pay to have it put in there. Collections in the MITAB library is one example I can remember off the top of my head. After the Mapserver / Autodesk thing though I am not sure how often the Oregon & SW Washington market of Comcast does this anymore last I heard they had stopped doing it due to that little fiasco but up until then they did it quite often. I am just saying not all large companies are bad about using Open Source and not giving back.

Re: Geospatial Code Sprinting

Author: Christopher Schmidt

I think the biggest thing that can be offered by many companies is computing resources. Via MetaCarta Labs, we provide over a half dozen tile sources -- from Blue Marble to Vmap0 to TIGER -- against a TileCache source. The result is that the world is now getting over 300,000 map tiles a day served by MetaCarta servers. The Telascience resources coming online via OSGeo are another example of this type of support: Landsat served from some set of high-powered servers which is not controlled by NASA, and therefore is not subject to the same beuracratic limitations. I think that fast server resources are the single thing that large companies have lots of that open source projects don't, and making them more available is the benefit that companies can offer.