On 10 Jan 2006, at 20:56, Riccardo wrote:
Hey,
On Tuesday, January 10, 2006, at 08:03 PM, Richard Frith-Macdonald
wrote:
On 10 Jan 2006, at 18:36, Marc Brünink wrote:
Hi,
is there any chance I'll be able to compile and run GNUstep in
64- bit mode on Solaris? Is the 64-bit mode stable enough to be
used on a productive machine?
quote out of the wiki:
Unfortunately, there are currently some problems with running
GNUstep configured under 64bit.
http://mediawiki.gnustep.org/index.php/
Installation_of_32bit_GNUstep_on_64bit_Linux
AFAIK nobody has tried using 64bit solaris.
I've just got 64bit intel (EMT64) system and installed Debian-
AMD64 on it. It's certainly not working fully for distributed
objects and for passing large structures in invocations, but
apart from that it seems OK so far. I guess it depends what you
I wasn't even able to get ffcall compile in 64bit mode... and I had
then problems elsewhere too,. It was a couple of months ago, but I
was far from running any application, so I can't even say if it is
stable...
FYI I've fixed a few bugs in base in CVS today and got rid of a lot
of warning messages.
I have distributed objects applications talking nicely.
I have invocations working well ... except for passing large
structures (something that most programs don't need anyway).
So if you can get a working copy of ffcall for 64bit solaris I think
you are in with a good chance of getting a usable system .. though
I'm sure there are lots of minor bugs to be tracked down. One place
that problems could lurk is with archiving/DO where the
initWithCoder: method tries to unarchive a long into a variable which
is actually only an int ... causing corruption of variables in
adjacent memory locations. I found one such bug today, and I guess
we need to go through all the initWithCoder: and encodeWithCode:
method implementations and check them.
_______________________________________________
Discuss-gnustep mailing list
[email protected]
http://lists.gnu.org/mailman/listinfo/discuss-gnustep
