Perl 6 Concurrency development

S17 is probably talking about "A language-based approach to unifying events and threads" by Peng Li and Steve Zdancewic, April 2006.

Interesting quotes

  • TimToady: philosophically, you need a monadish translator between the transactional view of processing and the event view
  • 09:12 < sorear> nine: I'd like to spend much more time talking about this with you.  I think S17 is insane in quite a few places
    09:14 < sorear> well, two mostly
    09:17 < sorear> 1. atomic code blocks as currently specified break the "you pay for what you use" rule, since every single Scalar access needs to check if there's a contend {} on the stack
    09:17 < sorear> I'd rather have my $x is txn;  but there may be a better way yet
    09:17 < sorear> 2. the correct (deadlock-free) way to handle asynchronous signals isn't control exceptions, it's starting a thread to handle them
  • 21:55 <@plobsing> nine: a good test would be to write a chameneous-redux benchmark and see that the whole thing didn't fall flat (like erlang does)

Powered by CiderCMS

Edit this page