Re: speculative instantiation and diagnostics

[Mark Mitchell <mark at codesourcery dot com>]

I've been worried about these problems for a long time, but I'm not quite
sure what to do.  The rules for template instantiation are very hard to
get right.

> Mark, it occurs to me that this is a similar requirement to the
> speculative parsing that the new parser must do, so a similar machinery
> should be used for both. Also the compiler's full of annoying checks for
> error_mark_node.  Starting from a clean slate, something like
> 	if (push_error_handler (&errorcatcher)) {
> 		handle error from do stuff
> 		either rethrow (errorcatcher)
> 		or return
> 	}
> 	... do stuff
> 	pop_error_handler ()
> This is something that could be implemented with setjmp/longjmp. And we'd
> need some way of squirrelling away a diagnostic, without the overhead of
> actually formatting it (until it really is emitted)
The first draft of the new parser did something a bit like this; it
actually queued up diagnostics until it was sure it wanted to emit
them.  That was complex and error-prone; in the new new parser, I've
thus far avoided having to do that.  EDG does not queue, so there
is an existence proof that it can be done.

--
Mark Mitchell                mark@codesourcery.com
CodeSourcery, LLC            http://www.codesourcery.com