Re: location lists revisited

[todd dot allen at ccur dot com (Todd Allen)]

> 
> As we decided at the last meeting, the only proposals which we plan 
> to consider are ones which either correct errors in the changes we 
> have made to Dwarf 2 which would make implementing support for Ada 
> more difficult, or proposals which remove impediments to implementing 
> support for Ada.
> 
> I see no difficulty with implementing this as a vendor extension,
> should you wish to do so.
> 
> Since this does not require any change to Dwarf to enable you to
> implement such a vendor extension, and there is nothing in the 
> existing standard which would impede your doing this, it would 
> appear that such a proposal would not fall into either of the
> categories we approved.
> 
> I believe that the committee anticipates that much of Ada support
> would be in the form of vendor extensions, for the many reasons
> which we gave at the committee meeting.   
> 

OK.  This is true.  It isn't necessary to modify DWARF2.1 to add support for
this, precisely because it can be done as a vendor extension.  I knew that
providing a means for describing this was generally useful and not just for
Ada, and that really was my impetus for suggesting a new proposal.  I'd hate
to see several different compiler vendors go off and implement different
vendor extensions to describe the same thing.

But if we're that late, then I'll simply let it drop, and go implement the
extension in our compilers.  

This still seems like a good forum to have discussed it.  This mailing list
seems to comprise most DWARF implementors in the world.  And this discussion
has been fruitful, I think.  If anyone has implemented or will implement
support for local copies of globals, and wants to discuss syncing up the
vendor extensions, I guess we can discuss it off-line.  Send me email.

-- 
Todd Allen
Concurrent Computer Corporation