Re: About to remove dwarfout.c

[DJ Delorie <dj at redhat dot com>]

> The solution in this case would be to use the new debug format.

IIRC we're working on it ;-)

The problem is that many of our djgpp-specific tools expect sdb format
and don't happen use binutils to parse it, so we can't just move to
something else without considerable effort.  We've been working on
supporting dwarf2, but it's taking time.

> It's not like this is a surprise change that zack has just sprung on
> everyone.

Except that this is the first I've noticed it.  My procmail looks for
keywords for parts of gcc I maintain and puts them in a higher
priority mailbox, so I noticed this one right away.

> It is not everyone's job to support every platform.  If someone removes a
> feature that everyone agrees should be removed (because it's no longer a
> featuer, it's heavily buggy, or very obsolete), then it should eb the
> responsibilty of that platform's maintainer to use the replacement
> provided.  Otherwise, there is no point in having maintainers for various
> targets except to fix bugs.  Part of platform maintenance is more than
> just bug fixing, it's keeping the code and features up to date with gcc
> proper.

In this case, though, proper procedure should be for the maintainers
to move their targets to the new feature *first*, and *then* remove
the old feature.  If Zack wants to remove the old feature before all
the targets are ready, then he's breaking those targets.