GSOC 2012

[Alexandre Courbot]

How about a project for a DUChain builder based on CLang? Provided
this is doable (I played a little bit with CLang and don't see any
reason why it would not be so far), there are many advantages:

- Full support for C/C++/C++11, languages features would not be mixed
as they currently are (i.e. a C project would raise errors on 'class'
and such),
- Better error reporting (and in sync with actual compiler messages
when people start using CLang instead of gcc)
- Would relief KDevelopers from having a maintain a language parser,
which is a huge cost in terms of manpower - would let them truly
concentrate on IDE aspects

Recently CLang has come to the point where it is capable of compiling
Qt and the Linux kernel, this says a lot about how close to being 100%
gcc-compatible it has become. CLang libraries provide a full AST of a
program in a reasonnable time and also have many configuration options
as to how parsing should be performed. With such a powerful tool at
hand I do not see the need to maintain one's own parser.

Alex.