Introduction at first.
I am a student, attempting at magister degree in "applied math" in Physical-Technical Institute of National Technical University of Ukraine “Kyiv Polytechnic Institute”. (don't be confused by double "institute", it a proper naming, although strange even for students).
When looking at the list of ideas suggested for google SoC I've found two of them that I'd like to be a part of.
First one - is fixing small memory leaks in ecpg, which is a certain variant, due to my experience with code digging, memory issues and memory profilers. It's quite a routine, but good practice task.
Second one is optional, due to it isn't marked as [E] task, but it strongly correlates with my knowledge.
It is implementing a faster CRC32 algorithm. I've read both discussion threads about it, and still consider myself probable of being involved in it.
My confidence in this fact is due to that fact that I am educated in algorithms estimation, codes, polynom operations and performance tests.
And here is the brief road-maps for this ideas.
The first task consists of:
* use profilers to identify automatically tracable leaks (1w)
* try to get rid of them (3w)
* try to find some complex leaks not found automatically (1w, just in case)
* test the changes (3w)
* cleanup and documentation (2w)
Second task, if acceptable, consists of mainly research and testing statistics for different variants and approaches.
There also can be a variant of early finishing idea with memory leaks cleanup, and contributing into the CRC task.
with regards,
Arsenii Bochechko