Re: [ajdt-dev] AJDT and ITDs

[Andrew Eisenberg <andrew@xxxxxxxxxxxx>]

On Wed, Jul 22, 2009 at 7:41 PM, Simone Gianni<simoneg@xxxxxxxxxx> wrote:
> As a (maybe) quick workaround for this problem (while waiting for THE
> solution :) ), AJDT could resolve to using the .java/.aj found in the source
> attachment of jars on the in path, if present.

Unfortunately, this still has the same problem in that AJDT's model is
borrowed from AspectJ, which is based off of the compiled code.  So,
there is currently no mechanism in place to generate a part of the
model without going through full compilation.

The best that we can do without being able to add our own custom
pointcut matcher is to get AspectJ to expand the scope of the model
that it produces.  Now, it generates a crosscutting model for a subset
of elements that it compiles (ie- all files from source, and all
aspects on the aspect path, but that's it.  In path elements are
excluded).  The reason for this is to keep the size of the
crosscutting model down and exclude binary elements.

Now, I don't know about the feasibility of including only those binary
elements that are the target of some aspect's ITD or pointcut.  This
might solve your problem.  However, Andy would have to speak to that.

--a

>
> This would greatly help a number of use cases for those who use open source
> libraries or internally developed libraries for which sources are present,
> but I have no idea if adding those plain java files to the model is easy or
> even possible without using all the infrastructure Eclipse offers with
> linked source folders.
>
> If I'm not wrong there are/were a few places in JDT and AJDT where
> programmatic parsing of code extracted for sources different than the .java
> in the source folders.
>
>