[omniORB] Incorrect documentation in .mk files

[Michael J. Donahue]

Quite right, not to mention that there is a mk file designed for use with the
gcc (i.e. powerpc_aix_4.2_egcs.mk)

FYI:  I'm told that IBM has commissioned Red Hat to update gcc to work fix the
multi-threaded exception handling issues on AIX 4.3.  This is strictly
unofficial, and I had to twist Red Hat's arm to get them to even admit they had
a contract with IBM.  That was last summer, and I haven't heard anything new
since.

- Mike





"Gary D. Duzan" <gdd0@gte.com> on 02/26/2001 12:21:10 PM

To:   Michael J. Donahue/MCLEOD@MCLEOD
cc:   omniorb-list@uk.research.att.com
Subject:  Re: [omniORB] Incorrect documentation in .mk files



In Message <862569FF.005E3704.00@smtp2.mcld.net> ,
   "Michael J. Donahue" <mdonahue@McLeodUSA.com> wrote:

=>
=>Duncan -
=>
=>Could you remove the following comments from the AIX xlC 5.0 mk files.  These
=>comments are
=>talking about the gcc compiler, but these mk files are for xlC 5.0.  Also,
even
=>if you were to try and
=>use these mk files per the comments for gcc, there would be far more changes
=>required to get gcc
=>to work.  Details of my request follow.


  True. Of course, the comments don't really hurt anything, but
they don't do any good, either, and arguably could be a distraction.
As it happens, there are bugs in gcc on AIX which prevent omniORB
from working properly. We have some patches to gcc 2.95.3 to get
it working, but it is a bit hackish in that special care has to be
taken in applications which may dynamically load C++ object modules
which use exceptions, as is the case with Python and omniORBpy.
Unfortunately, these changes aren't likely to be accepted by the
gcc maintainers, and the real fix which should be in 3.0 won't be
out for quite a while.  If anyone is particularly interested in
the changes we might consider putting a patchkit together. Let me
know.

                         Gary Duzan
                         Verizon Laboratories