Re: Handle java_demangling?

[Hans-Peter Nilsson <hans-peter dot nilsson at axis dot com>]

> Date: Thu, 16 Nov 2000 18:56:30 -0500
> From: DJ Delorie <dj@delorie.com>

> > Well, is the master copy in gcc?

Most definitely.

> Yes, but cplus-dem.c is being maintained by someone other than me (as
> libiberty maintainer).  I don't know exactly what their policies are,
> but they don't follow the usual libiberty policies.  Stuff seems to
> get done anyway ;-)

I would like to think Dan Berlin and I would follow "libiberty
policies", but I don't know exactly what you refer to, so I
can't say.  Recent breakage indicates some kind of not-following
policy, though. :-( Sorry.  Do we misbehave in other ways too?
Or do you just refer to maintaining cplus-dem.c a bit different
from other libiberty sources?

IIRC, the change in policy to maintain cplus-dem.c with a few
extra hands *in addition to* other libiberty maintainers, was
done too speed up getting demangler fixes done -- but I'm not
sure we are successful.  Note that cp-demangle.c has another
maintainer.  Perhaps we should actually say this in the various
MAINTAINERS files...

> However, I do ensure that binutils/gdb get updated when gcc gets
> updated, either by nagging the contributor (preferred) or merging it
> myself.

It would be nice to have a few testcases for Ada and Java
demangling.  Not that simply adding testcases would have caught
the seemingly trivial missing "case" that H.J. found; it seems
to happen only when demangling stdin.

I will bootstrap GCC with H.J.:s patch and run the GCC and
binutils testsuites.  If it passes, I'll commit to GCC and
binutils.

Thanks.

brgds, H-P