I returned home two days ago from the 2010 KDE Imaging sprint. It was awesome because:
- It was fun to “meet the people behind the email addresses”,
- I learned some new stuff from seeing how the others code,
- I’d never been outside of India before.
I arrived in Marseille sometime before noon, after a journey of more than 24 hours, barely catching about 3 hours of sleep (I still don’t get how people can sleep in an airplane). At the airport I met fellow digiKam contributor and SoK student Kunal, and managed to catch a bus to Aix en Provence, which is approximately a 20-minute ride from the airport. Aix is Gilles Caulier’s home city, and, I must say, a beautiful one.
Most of us arrived a day earlier, and in the evening, Gilles invited us to a digiKam talk that he was to give in a local LUG meeting. In the same place where we would code the next day. It was fun chatting with the LUG folk (some of them knew English).
I had to crash on bed soon that night, I was sort of starved of sleep.
The next day, I woke up early and walked around the hotel’s neighborhood, taking pictures of whatever caught my eye. There was this really nice cathedral there, just a five-minute walk from the hotel:
Gabriel arrived that morning, and then we proceeded to the Maison des Associations Tavan, which is the hall where we would work. We just called it “Tavan”.
During the coding, Marcel showed me his new revamped implementation of the face interface in digiKam, and we discussed with Gilles some of the possible reasons behind several crashy bugs in the face scan process. Marcel had committed the face thumbnailer delegate he’d been working on since a while, and I and Gilles played around with it for a bit.
We had a fair amount of false positives in the face detection, and while this accuracy can be improved in libface, there will always be some false positives, so we needed a “reject button” overlay on each face thumbnail to be able to dismiss these.
I set to work on that, with help from Marcel. I wasn’t able to finish it due to some bugs in my code, but it looks like Marcel has fixed it now.
Michael had brought a really nice device that he had assembled and programmed himself :
Meanwhile, we discussed the feasibility of having arbitrary (as in polygonal/freeform closed curve) region tagging in digiKam, which seems an enticing idea. Some other future improvements such as having truly multithreaded batch tasks (like fingerprinting), and having a unified task manager for controlling these were discussed.
Gabriel and Michael worked on the polishing stuff in the reverse geocoding project, and Martin continued working on the image versioning. This was the first time I saw Gabriel’s project in action. Kunal continued with QtScript integration, and we discussed interesting use-cases of this very useful and enabling project. So I guess Hot New Stuff ™ is coming to digiKam soon! So we’ve all been working on our projects post-GSoC too ;-) Andreas Huggel was there for Exiv2 work, and Laurent Espitalier for database interop with other applications.
I’m looking forward to working for a long time with the digiKam team. And hopefully I’ll be able to attend the next sprint too.