As noted on the project ideas page, keeping this open and on the xapian-devel mailing list. On 30 Mar 2011, at 05:34, Harmandeep Singh Pall wrote:> I am very much interested to work for your org as a Google Summer of Code student. I am interested to work on the project that intends to improve Xapian's Python bindings ( http://trac.xapian.org/wiki/GSoCProjectIdeas#PythonBindingsImprovements ). I would like to know if you require students with past experience or patch submission for bug fix. I would also like to know if it has to be swig based implementation or hand coded and other strict specifications you would like me to know, if any.I doubt we'd be comfortable with the idea of replacing a SWIG-based binding with a different one without a very robust justification, because it would dramatically increase future maintenance costs (and since Python is one of our most important languages to support, that could prevent our merging in such work). Beyond that, we're looking for any and all changes that improve the experience of developing with Xapian in Python. We certainly don't require students to have past experience of generating patches or working with open source projects; we do however ask you to list any such experience in the background information of our application template. Best, James -- James Aylett talktorex.co.uk - xapian.org - devfort.com
On Wed, Mar 30, 2011 at 12:04:06PM +0100, James Aylett wrote:> I doubt we'd be comfortable with the idea of replacing a SWIG-based > binding with a different one without a very robust justification, > because it would dramatically increase future maintenance costs (and > since Python is one of our most important languages to support, that > could prevent our merging in such work). Beyond that, we're looking > for any and all changes that improve the experience of developing with > Xapian in Python.I'd be very resistant to moving to hand-coded bindings for a language well supported by SWIG. We've already been there with Java and Perl, and it just means that new API features don't get wrapped promptly. If there's another tool which can automate wrapping the simple cases and help wrap the more complex ones, then I'm happy to consider that. We get some minor benefits from using a common tool for several languages, but a better tool could easily outweigh these. Cheers, Olly