user_Jamesday wrote:
I don't think it's wise to have four parsers doing actually the same. <<
On the server side it seems that speed is a desired feature, so C or C++ seems to make sense as a preferred language. They also have the advantage of being very widely understood. Broad understanding probably favors C or simple C++ (simple C++ meaning without really fancy use of classes and inheritance). Hopefully either of those would allow a lot of people to get involved in optimizing and reusing the code.
Well, I did use classes, but we do so already in PHP. I didn't do "fancy" stuff like multiple inheritance, though. IMHO the structure of the code is close enough to the current script that people who know MediaWiki well should be able to read most of the code, even if they didn't work with C++ before.
Example : What might this do?
TUCS t = readSomeFreakingText() ; VTUCS v ; t.explode ( "\n" , v ) ;
(Hint: TUCS is my string type, VTUCS is an array of such...)
Magnus