[Mono-devel-list] mcs patch for precise location handling

[Allan Edwards]

Does anyone in the mono project keep a comprehensive list of what is 
implemented in Windows.Forms and what is not.  I have been playing with 
porting Infragistics to Linux and it is just not clear what will run and 
what won't.

Thanks,
Allan

Atsushi Eno wrote:
> Hi Marek,
> 
> Thanks for the comments.
> 
>> That is massive ;-)
> 
> 
> Well, I can split the patch into maybe two parts: 1) make changes
> to Location, Report and tokenizer, and 2) make changes to parser
> (lexer.Location elimination). But 1) won't make sense as a
> standalone patch. And in fact most of the changes are in 2).
> 
>> In the most cases we don't need to pass Location as extra parameter now.
>> It would be nice to remove also this redundant casts and arguments.
>> For example.
>>
>> $$ = new ComposedCast ((Expression) $1, "*", ((Expression) $1).Location);
>>
>> or
>> $$ = new Invocation ((Expression) $1, args, ((Expression) $1).Location);
>>
>> replaced by
>>
>> $$ = new ComposedCast ((Expression) $1, "*");
>>
>> $$ = new Invocation ((Expression) $1, args)
> 
> 
> Yeah, I totally agree.
> 
> So, based on your suggestion, I created further patch. Now it's
> much more massive than before (hehe).
> It recuded Location argument in several expressions and statements.
> For members in iterators (MoveNext() and so on), it reuses the
> same Location of the corresponding iterator itself.
> http://monkey.workarea.jp/tmp/20050627/precise-location-20050628.diff
> 
> Total memory allocated decreased a bit: 77618 KB
> http://monkey.workarea.jp/tmp/20050627/perf-new3.txt
> 
> Additionally, I also added the fix for bug #57047 as Hari suggested
> in tokenizer, parser and Attribute class so that Attribute accepts
> "nameEscaped" bool parameter which is passed from tokenizer.
> 
> I think this patch can be delayed for review after the first one.
> 
> Atsushi Eno
> _______________________________________________
> Mono-devel-list mailing list
> Mono-devel-list at lists.ximian.com
> http://lists.ximian.com/mailman/listinfo/mono-devel-list
> 
> .
>