Re: Patch to add BeOS/x86 support

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

  In message <200001112248.OAA13094@casey.cygnus.com>you write:
  > The BeOS is an odd puppy, gratuitously incompatible with Unix
  > conventions in a number of ways.  Several of these ways are new to
  > gcc, so they have been parameterized.
You're being awfully nice in how you describe BeOS's "features" :-)

  > Older versions of BeOS had a 256K stack, which wasn't anywhere near
  > enough to run gcc using the normal alloca, so it was necessary to
  > force use of the C alloca.  Newer versions have a 1M stack, which is
  > still pretty iffy, though less of a problem since gcc's alloca usage
  > has decreased.
Yea.  I would claim that if we're allocating a meg of stack, then something
is probably wrong -- with the possible exception of absurdly deep call chains.

  > The folks at Be wanted 'gcc foo.c' to produce 'foo' rather than
  > 'a.out'; this patch implements that change for the BeOS port.  How do
  > people feel about this inconsistency?
Sigh.  I guess we'll live with it, though I'm not particularly happy about it.
jeff