I have been able to invest some time on the JParser code this last two
weeks and it has come to a fairly stable state. The branch isn't
stable enough to get merged back to the trunk, but it can be used (as
a beta version, of course) by end-users. Previous developer previews
were most useful for non-win32 users, who didn't have any parsing
support. But, because of non-standard declarations support, this one
will also be appreciated by win32 users.
Since December 6th, I have mostly worked on code layout rules, but
there were improvements in some other areas too. In my opinion, the
most special one is the treatment given to non-standard code, like the
"foreign" declarations on our old parser code and the pre-processor
directives on darcs' code. Now the parser is able to ignore those
non-standard declarations and normally parse the code after them. I
think this isn't supported by the old win32 parser (I can't say for
sure, since I couldn't really test it).
A ModuleBuilder component was added to the parser too. This is the
first step towards a domain object model and allows the parser to
group various function matches in one unique function binding. The
most noticeable difference is seen on the outline view. Without this
builder, the outline for the following program
module Factorial where
fac 0 =3D 1
fac n =3D n * (fat (n - 1))
now it looks like
Some changes were also made to make the code compliant to the latest
platform milestone (Eclipse 3.2M4). The plugin runs almost smoothly
(there may still be some error messages) on a 3.2M4 or 3.1.1 platform.
Comments are much welcome.