Hi David and all!
I just want to say: kudos.
I personally have no real need right now for 64 bits (I think?) but I recognize grunt work when I see it. :)
regards, Göran
PS. Btw, I have two Ultra5's running Debian in my attic, not sure if those can be used for any interesting tests or so.
On Nov 15, 2007 2:29 PM, goran@krampe.se wrote:
PS. Btw, I have two Ultra5's running Debian in my attic, not sure if those can be used for any interesting tests or so.
I have an Ultra-60 (64-bit, big-endian) and a DEC Alpha (64-bit, little-endian) I'm keen to help too, but where does one start?
Andrew
On Fri, Nov 16, 2007 at 08:18:13PM +0200, Andrew Gaylard wrote:
On Nov 15, 2007 2:29 PM, goran@krampe.se wrote:
PS. Btw, I have two Ultra5's running Debian in my attic, not sure if those can be used for any interesting tests or so.
I have an Ultra-60 (64-bit, big-endian) and a DEC Alpha (64-bit, little-endian) I'm keen to help too, but where does one start?
It certainly couldn't hurt to try running some different machine architectures. My "64 bit server" machine is a tiny laptop running Linux. Performance testing on some real servers might be interesting, particularly for the 64-bit VM/32-bit image combination, which would be used for Seaside servers and such.
You should be able to build 64-bit VMs for 32-bit image on those machines without too much difficulty. For 64-bit images, it's probably easiest if I send you a working image, because the original 64-bit image from squeakvm.org cannot be run on a current VM (come to think of it, the procedure for working around that gotcha would be worth documenting one of these days).
In addition to updating the plugins, there are probably lots of issues and bugs that will show up in the 64-bit image if you run some of the SUnit tests. I'm not sure if any of these are worth working on at this point, but identifying some of the problem areas might be helpful. I suspect that a large proportion of things that appear broken in the tests will turn out to be problems in the tests themselves, but getting that right is just as important as fixing any problems in the image per se.
At some point we should try to convince Tim to copy his archive of VMMaker-tpr.00.mcz through VMMaker-tpr.nn.mcz over to SqueakSource before he gets run over by a truck and the change history is lost to the ages.
Dave
vm-dev@lists.squeakfoundation.org