On Feb 21, 12:21 am, Alex Ghitza <aghi...@gmail.com> wrote:
Hi Alex,
> I've recently (yesterday) turned my work MacBook into a dual-boot MacOSX and
> Archlinux machine. I built 3.3.rc3 from scratch under Linux, without
> problems and with all tests passing. I updated to 3.3.rc3 under OSX,
> without problems except one small doctest failure due to the fact that I
> forgot I had a patch applied before the update.
Ok.
> So all looks good. Slightly off topic though: "make test" takes 4400
> seconds under OSX, and only 2700 seconds under Linux. In both cases, I was
> only running Firefox with Gmail in it. That's a serious difference!
Yes, it is a known issue (at least I have known about it for a long
time) and it is even more pronounced on OSX 10.4 than 10.5 compared to
Linux. This has to do with the rather sluggish linker on OSX and I
believe can mostly be blamed on Apple supporting things like symbol
lookup at runtime.
I have heard that OSX 10.6 is supposed to be better here - allegedly
even significantly, but I have not tested on any 10.6 box since I
haven't gotten around to install the preview and fix the build issues
in Sage that inevitably crop up :)
To resolve this problem I believe it would be easiest to use
pyprocessing for doctesting, i.e. initialize a Sage session and then
fork off an instance and run the doctest with it. Something similar
would come in very handy for the notebook since that way one could
have a master Sage process and for each worksheet creation we would
just have to fork an instance which should be substantially cheaper
than a startup. Gary has done some experiments in that direction, but
AFAIK his work is neither in trac nor is he actively working on this.
One issue here is dealing with sockets for example, but if we had a
special "just enough Sage" mode that would avoid this and/or potential
similar issue I would be very glad to see it merged into Sage :)
> Best,
> Alex
Cheers,
Michael
--~--~---------~--~----~------------~-------~--~----~
To post to this group, send email to sage-devel@googlegroups.com
To unsubscribe from this group, send email to
sage-devel-unsubscr...@googlegroups.com
For more options, visit this group at http://groups.google.com/group/sage-devel
URLs: http://www.sagemath.org
-~----------~----~----~----~------~----~------~--~---
