wierd problem about egcs1.1.1 win32 binary release

[Jeffrey A Law]

  In message < E10SGx8-00052e-00@roboto.devolution.com >you write:
  > >   > -Os doesn't lower code size much in my experience, and does result in
  > >   > slower code on the 686.
  > > Testcases please (for -Os increasing codesize)?  Without testcases we
  > > can't actually fix the problem.
  > 
  > Oh, it didn't increase codesize, it just didn't lower it by much.
  > The code in question was really hairy template code, so I'm not concerned. 
  > :)
Ah.  It's still a work in progress anyway.  It seemed kind of silly to keep
complaining that we couldn't do code size opts because we didn't have -Os.  So
we added -Os and now we iterate on changes that actually improve size.

egcs-1.1 did very little with -Os.  egcs-1.2 does more, and I expect we'll
continue to add more space opts over time.


jeff