ASPN ActiveState Programmer Network
ActiveState
/ Home / Perl / PHP / Python / Tcl / XSLT /
/ Safari / My ASPN /
Cookbooks | Documentation | Mailing Lists | Modules | News Feeds | Products | User Groups


Recent Messages
List Archives
About the List
List Leaders
Subscription Options

View Subscriptions
Help

View by Topic
ActiveState
.NET Framework
Open Source
Perl
PHP
Python
Tcl
Web Services
XML & XSLT

View by Category
Database
General
SOAP
System Administration
Tools
User Interfaces
Web Programming
XML Programming


MyASPN >> Mail Archive >> perl6-language
perl6-language
Re: [perl #36250] [TODO] STM threading design
by Larry Wall other posts by this author
Jun 11 2005 10:39PM messages near this date
Re: How much do we close over? | Re: q:e/.../ as a short cut for eval q/.../
On Sat, Jun 11, 2005 at 02:23:10PM -0700, Chip Salzenberg wrote:
: STM is Software Transactional Memory.  It's wicked cool.  Really.
: Google it.  You'll be glad you did.

I'm particularly interested by the scalability of

    http://www.cambridge.intel-research.net/~rennals/faststm.html

(There's an implementation on Source Forge as well.)

: BTW, if you're wondering how this can work efficiently in Parrot:
: Remember that threading contention only applies to those variables
: which are marked 'shared'.  So the STM system need only look at their
: accesses transactioned.

On the other hand, some of the parallelizable primitives we're
building into Perl 6 (junctions, hyperops, and especially pipes) may
well induce more accidental "sharing" than appears on the surface
(think of a lexical that is visible to both ends of a pipeline).
So STM is probably a good direction to be moving in any case.

And if we need special syntax to keep such accidental sharing under
control, now would be a good time to discover that rather than later.

[Note: followups directed to p6l by default.]

Larry

Privacy Policy | Email Opt-out | Feedback | Syndication
© ActiveState Software Inc. All rights reserved