[portland] oscon recap (was: Planning for the August Meeting)

kirby urner kirby.urner at gmail.com
Sat Aug 4 21:00:06 CEST 2007

On 8/4/07, Kevin Turner <kevin at freegeek.org> wrote:
> On Tue, 2007-07-31 at 21:24 -0700, Mark Gross wrote:
> > The multi-core python wasn't about python so much as it was about the
> > cell in a PS-3.  perhaps I was tiered and missed something, but you can
> > check what materials that have been posted so far:
> > http://conferences.oreillynet.com/pub/w/58/presentations.html

I did go to talks on parallelism, though not the one above.

Intel has released a new open source C++ compiler that's designed
for multi-core.  Haskell team @ Microsoft is pushing an ACID-based
design cribbed from data base world (my world):  atomic tasking
with rollback and conditional switching.  In other words, you could
have like a Python try-scope with rollback in case of an exception
of type Not_Parallel (e.g. it'd break some synchronization rule) --
in which case try something else.

> > The Python 3000 presentation by Guido was kind of dull.  It was about
> > how to get ready for the switch over and how you should port to 2.6
> > (when its out) and always use the auto-port tool to make the code 3.0
> > ready.
> Jeff was pretty troubled with this one.  Jeff, has anyone you talked to
> since then convinced you that there's a reason Guido is breaking the
> language?

Which Jeff is that?  Deliberate breakage of the language in going from
2.x to 3.x has been in the cards for like seven years.  The question isn't
if, but how.  That the answers are starting to sound dull is a good sign:
means it's already starting to happen.


More information about the Portland mailing list