Re: egcs-19980906 patches for bootstrap failures ...

[Jeffrey A Law <law at cygnus dot com>]

  In message <199809080352.XAA05507@caip.rutgers.edu>you write:
  > 	Hmm, I had assumed one liners which were necessary to get egcs
  > to compile should be exempt from paperwork, regardless of whether they
  > were for gcc or java.  Now I'll just post the errors and let you fix
  > them instead.  :-/
Until I hear back from Java folks and the lawyers I'm taking a *very*
conservative stance on Java patches.  There's no sense in giving a
certain company any daylight in regards to possible litigation over
our Java code.  Hopefully we'll have the necessary forms for Java
contributions ready soon.

  > 	BTW, it seems that the java/ dir is being built in stage1
  > for some reason.  This is a build bug, right?
Hmmm, it's some subcomponent of the java tree, but not the compiler
itself.  Not exactly sure what the one binaries does though :-0

  >  > Seems to me that "no_argument" is a namespace violation from getopt.h
  >  > and that getopt.h should be fixed instead.
  >  > 
  > 
  > 	True.  But is it practical to change these macros in getopt.h
  > considering how ubiquitous gnu getopt is? In the mean time, egcs won't
  > compile.  What would you like me to do?
Yes it is practical since we find the one in our own source tree.

We should (of course) forward those patches back to the FSF.  Actually
we should check if the FSF already has this bug fixed in the latest
and greatest getopt code.


jeff