Re: Still a lot of C++ files getting "fixed"

[Zack Weinberg <zack at wolery dot cumb dot org>]

On Tue, Feb 01, 2000 at 12:02:20PM -0800, Bruce Korb wrote:
> Zack Weinberg wrote:
> > A bit more than just that can go away...  There's also fixcpp,
> > fixinc-nt.sed, and a bit of Makefile cruft (patch appended).
> 
> Well, I haven't done anything to address NT.
> I know it has pretentions of being POSIX,
> but I did not feel comfortable playing around
> with that environment.  

fixinc-nt.sed isn't referenced by anything.  The NT fixincludes script
is over in fixinc/fixinc.winnt.

Note that all the 'winnt' target files appear to be massively out of
date and could probably be scrapped without anyone noticing.  We have
mingw32, after all.

Looks like just-fixinc should be deleted too.

> > Note I'm still waiting for review on the last round of corrections to
> > the new fixincludes, and Kaveh's patch to not rerun fixinc each pass
> > hasn't been approved either.  (The patch below includes Kaveh's patch.)
> 
> Uh, oh.  I was out of town for six days and found
> hundreds of emails upon my return.  Busy days.
> Please give me some titles that I should look at
> so I can be sure I can pull them out of the gcc-bugs & patches
> archives....Kaveh's patch is clearly needed and looks reasonable
> enough to me

Patch I'm blocked on:

http://gcc.gnu.org/ml/gcc-patches/2000-01/msg01095.html

Related discussion:

http://gcc.gnu.org/ml/gcc-bugs/2000-01/msg00878.html
http://gcc.gnu.org/ml/gcc-bugs/2000-01/msg00637.html

and subsequent threads.

Executive summary: I changed cpplib to accept // comments and #endif
FOO in system headers, which means the else_endif_label and
no_double_slash fixes are redundant.  We should theoretically have a
fix for "//*" -> "/ /*" but I'm going to skip it until it actually
bites someone.  Also, machine_name didn't deal properly with having
nothing to do (e.g. AIX's predefined identifiers are all in the
reserved namespace).

zw