On Tue, 2006-03-07 at 11:27 +0100, Vittorio Rigamonti wrote: > Hi all, > > about this problem I have two candidate ways of proceeding: > > 1) starting from the OMG IDL spec. > copy the grammar rules into a text file > with some vim substitutions produce a javacc jj file > generate the lexical analyzer with javacc > develop a java idl2java compiler > Yeah, I'd go the java route, since that makes it easier to use features like reflection, serialization, etc, if they turn out to make sense. > I'm not sure if there are licensing issues about the following points: > > copying from the OMG specs (but I think this is safe) Implementing the same grammar is fine. > using as Open Source a *product* of javacc > Given javacc's weird legal status, I'd recommend not using a tool as part of our build process for which we can't be reasonably sure that it is free software. If someone is working in a nuclear facility, they should still be able to regenerate the idl2java compiler without violating some tool's license. ;) > > and what abount ANTLR? (just discovered!) http://www.antlr.org/license.html > > Do you have opinions? I'd go for antlr, since it is pretty cool, and being used in gjdoc already. cheers, dalibor topic
