Re: PATCH: new flag to say when things are instantiated

[Gabriel Dos Reis <gdr at codesourcery dot com>]

Brendan Kehoe <brendan@zen.org> writes:

| > I think the diagnostics should be more descriptive, insisting on the
| > fact that an instanciation is implicit or explicit.  For function
| > templates and static data member of class templates, we should
| > distinguish between "declaration instanciation" and "definition
| > instanciation".   According to standard terminology, it is technically
| > incorrect to speak of "variable instanciation".
| 
| I agree that those things could be used to further enhance it or make 
| its use of words more technically accurate.  However, I also don't know 
| when I'll have time to do that further work on the change, and wanted to 
| get at least a start in place rather than let these changes sit unseen 
| in my tree for another six months.  (As they have for the last six.)
| 
| Make sense?

I see.  As far as the option spelling is concerned, I would like to
point out that GCC tends to have tons of options (for good things) and
spelling them out completely seems (to me) to be better than
contracting them.  We've started following the logic of complete
spelling.  I understand that the fear that we'll end up with
riduciluous command lines but:

	1) people have learnt to use tools like Makefile or such to
	   automate builds;
	2) there are competitive popular compilers (should I say which
	   perform better than GCC) which demonstrated that that is
	   not a problem for users as far as the request services
	   worth it.

Apart from that, the diagnostic part is fine with me; you might need
Mark's approval for the rest.  Would you mind changing the option
spelling?   

Once in, I'll see what to do with it :-)

Thanks,

-- Gaby
CodeSourcery, LLC                       http://www.codesourcery.com