Detect re-throws outside of a catch block.

Detect re-throws outside of a catch block.

R=johnlenz
Committed: https://code.google.com/p/dart/source/detail?r=407

[John Lenz, 2011-10-12 21:25:07]

LGTM

[mmendez, 2011-10-12 21:25:39]

I don't think that we should be checking this in the parser since it is not a
syntax error.  Also, we should not mutate the AST by injecting a string.

http://codereview.chromium.org/8247007/diff/1/compiler/java/com/google/dart/c...
File compiler/java/com/google/dart/compiler/parser/DartParser.java (right):

http://codereview.chromium.org/8247007/diff/1/compiler/java/com/google/dart/c...
compiler/java/com/google/dart/compiler/parser/DartParser.java:2500: exception =
DartStringLiteral.get("Invalid re-throw");
Is this string literal necessary here or even correct?

This will jackup the IDE because the string will now appear in the AST.

[jat, 2011-10-12 22:19:49]

I still do not like the idea of moving the check out of the parser -- this is an
error that is noticed from the static structure of the program, and has nothing
to do with either resolving identifiers or analyzing types -- of the two,
TypeAnalyzer seemed the slightly lesser evil.

[mmendez, 2011-10-13 15:48:28]

LGTM

I'm not crazy about doing that in the TypeAnalyzer either.  We need to revisit
this structure in general, which should take place within the next week or so.

Could you follow up with a test that ensure that we rethrow the correct
exception instance?  I'm not sure if we have a test for that already or not.


On 2011/10/12 22:19:49, jat wrote:
> I still do not like the idea of moving the check out of the parser -- this is
an
> error that is noticed from the static structure of the program, and has
nothing
> to do with either resolving identifiers or analyzing types -- of the two,
> TypeAnalyzer seemed the slightly lesser evil.

[jat, 2011-10-13 18:39:36]

On 2011/10/13 15:48:28, mmendez wrote:
> LGTM

Committed at r407.

> I'm not crazy about doing that in the TypeAnalyzer either.  We need to revisit
> this structure in general, which should take place within the next week or so.

Ok.  I still don't see why this is any different than, for example, giving an
error on a static abstract method -- both are discoverable entirely by parsing
without making any interpretation of symbols or types.
 
> Could you follow up with a test that ensure that we rethrow the correct
> exception instance?  I'm not sure if we have a test for that already or not.

Yes, in a separate CL.

[jat, 2011-10-14 18:06:42]

On 2011/10/13 18:39:36, jat wrote:
> On 2011/10/13 15:48:28, mmendez wrote:
> > Could you follow up with a test that ensure that we rethrow the correct
> > exception instance?  I'm not sure if we have a test for that already or not.
> 
> Yes, in a separate CL.

RethrowTest added in r448.