> That is exactly what I was getting at in the earlier discussion --
> conversion to a source name (which this essentially is, just
> separated at the dots for you) is lossy.
Right, that is primarily what my patch is attempting to fix. This
was one the spot (...well, deferred binding aside) I had
On Thu, Sep 8, 2011 at 1:36 PM, Stephen Haberman wrote:
> For some reason when I put it in, I thought it would be easy to come
> back later and fix. But that is not turning out to be the case.
>
> I really expected qualifiedRef to be a "regular" compound name, e.g.
> [foo, Bar$Zaz], but instead i
> http://gwt-code-reviews.appspot.com/1542804/
Well. I'm stuck on this TODO:
https://github.com/stephenh/scalagwt-gwt/blob/lessnamemangling/dev/core/src/com/google/gwt/dev/javac/CompilationStateBuilder.java#L106
For some reason when I put it in, I thought it would be easy to come
back later and
> We chatted, and Ray C is up for reviewing this, but I'm afraid it'll
> be the usual maybe-this-week drill.
No problem.
I uploaded the patch as:
http://gwt-code-reviews.appspot.com/1542804/
Thanks!
- Stephen
--
http://groups.google.com/group/Google-Web-Toolkit-Contributors
Reviewers: cromwellian,
Please review this at http://gwt-code-reviews.appspot.com/1542804/
Affected files:
M dev/core/src/com/google/gwt/core/ext/typeinfo/JClassType.java
M dev/core/src/com/google/gwt/core/ext/typeinfo/TypeOracle.java
M dev/core/src/com/google/gwt/dev/asm/Type.java
M d
Thanks Steve!
We chatted, and Ray C is up for reviewing this, but I'm afraid it'll be the
usual maybe-this-week drill.
On Fri Sep 02 13:06:44 GMT-700 2011, Stephen Haberman wrote:
>
> > I'd prefer this problem to be addressed in separate CL independent
> > from of our effort.
>
> Here is a comm
> I'd prefer this problem to be addressed in separate CL independent
> from of our effort.
Here is a commit with just the name mangling. If there are any
volunteers from the GWT team to review it, I'll promote it to
an issue.
https://github.com/stephenh/scalagwt-gwt/commit/5fb1f9717424b8c604
On 1 September 2011 20:52, Stephen Haberman wrote:
>
> So, in summary, yeah, this is a lot of changes. But a lot of name
> mangling is gone and we've been using this in scalagwt with more success
> than we had before.
>
> I'd be glad to answer any questions, comments, concerns about the
> approach.
Hi,
One of the challenges we hit with scalagwt was source <->
internal/binary name conversion being hueristic. Greg started a
prior discussion about this, with the summary basically being "yeah,
it's a heuristic, and yeah, it's a problem".
To get scalagwt working in the least-hacky-way possible,
