On 30 Dec 2009, at 18:22, Tudor Girba wrote:
Hi Johan,
I am not really considering this as a parser, as the plugin would directly rely on the Eclipse model of the underlying code, which solves the parsing issues for us. (But certainly there will be other issues :-/ )
Sure, the parsing part will be solved, but you will still have to deal with the evolution of the AST and the interpretation of the AST, which is not trivial at all.
As I said, other issues ;-)
It's cool to see that there has been work done before on that! So my question is what we can learn of these efforts, what is the story here? Is there any clear picture on what was wrong with these multiple plugins and how each successive attempt solved/failed to solve issues ?
Well, probably Moose Brewer is the most advanced and up to date solution, being also the latest effort.
OK, so that could be a starting point.
The main problem is that people have just left or shifted their interest. The Moose Brewer and j2moose plugins were started by students and as they left, nobody picked them up. jfamix was started by Frank Buchli as a side project, but as he got more busy with work he shifted away.
In the end, nobody picked these projects up and they all remained unmaintained.
So, all in all, the main lesson for me is that this kind of projects should not be dealt with through a one-time effort.
True, to keep it up-to-date it needs to follow the evolution of the language, which with Java is a considerable issue :-/ Sounds like a yearly student project ;-)
-- Johan Fabry jfabry@dcc.uchile.cl - http://dcc.uchile.cl/~jfabry PLEIAD Lab - Computer Science Department (DCC) - University of Chile