Re: Version 2.1.pre1 of the crypt extension

[Jeffrey A Law <law@hurl.cygnus.com>]

  In message <199902261715.MAA05330@blastula.phys.columbia.edu>you write:
  > >All the egcs that I have compiled so far (none with glibc2.1 though),
  > >including one for mips-irix6, show this `problem', while SGI cc
  > >passes.  So, is this really an egcs bug and has nothing to do with
  > >glibc ?  If so, I'd like to report it as a bug to egcs-bugs, because
  > >egcs-1.1.2pre2 still shows the problem.
  > 
  > We've got this fixed now in development egcs.  Jeff, you might want to
  > consider putting Kaveh's HOST_WIDEST_INT stuff into 1.1.2.  The above
  > bug is causing a real problem with libc 2.1.
Someone would have to build a patch relative to the release branch -- an
awful lot of stuff has been changing in cpp land and I'm not particularly
comfortable with extracting cpp changes out of the mainline tree and trying
to backport them to the branch.

To give you some clue about how much has changed, egcs-1.1. branched at
revision 1.30 in cccp.c.  We're now at revision 1.61.  Yow!

Some of the changes where relatively small, others where rather large.

Our time may be better spent documenting this failure (since I believe it
has always been broken in cccp), getting egcs-1.1.2 out the door, then
working on egcs-1.2 (which has lots of fixes in this area).

I'll consider it if someone can get me a patch relatively soon.

I know we'd need the introduction of HOST_WIDEST_INT and 
s/HOST_WIDE_INT/HOST_WIDEST_INT/g in cccp.c (others?, we may also need
s/unsigned_HOST_WIDE_INT/unsigned HOST_WIDEST_INT/g

And some fixes to the actual functions which do arithmetic in cccp.c
(left_shift, right_shift, etc).




jeff