Merge from pch branch; call for testers

[Tim_Josling@national.com.au]

Geoff Keating <geoffk@geoffk.org>
05/06/2002 14:20
Please respond to Geoff Keating

To: Tim_Josling@national.com.au
cc: gcc@gcc.gnu.org
Subject: Re: Merge from pch branch; call for testers


>> I guess I can reverse engineer what is required but a few hints or even
>> some documentation would be very helpful.

> How about looking at the manual?  It's node "Type Information" in
> gccint.info.

I had scanned the doc directory but none of the files or keywords I tried
were found. The info build actually fails at the moment with mismatched
braces in invoke.texi but gt.texi has the information I need.

> I do not recommend reverse-engineering.  The current gengtype.c is not
> final and many parts of it are intended to be replaced in the future.
> If an example is desired, I'd suggest looking at the changes to the
> Fortran frontend, which is simple, and the C++ frontend, which isn't.

>> It appears I will need to patch gengtype.c to add treelang to the hard
>> coded list of languages (lang_names) included therein. Is there any
>> alternative to this? Normally the determination of languages is by a
>> discovery process rather than hard coding in sundry files.

> Please feel free to suggest a patch.

OK.

The documentation doesn't mention the need to update gengtype.c (actually
in several places). Have I misunderstood?

> Geoffrey Keating <geoffk@geoffk.org> <geoffk@redhat.com>