I came across a thread from Jul '08 which seems to be the definitive on handling side-effects within transactions - http://groups.google.co.za/group/clojure/browse_thread/thread/d645d77a8b51f01/667e833c1ea381d7
Adrian. On Wed, Apr 1, 2009 at 9:24 AM, Timothy Pratley <timothyprat...@gmail.com> wrote: > > Hi Korny, > > > That is a very interesting question to me, specifically this part: > >> how do I stop parallel >> changes to two unrelated structures in the world from causing transaction >> retries? > > Actually I don't think you can ever rule out retries if you are using > a ref (or atom) and more than one thread to access it. For me this > poses a problem because my 'transaction' has side-effects. I have a > function which modifies the state, but it also reports the > modification (reporting should only happen once). I can think of two > solutions: > (1) Use an agent instead - changes are coordinated on a queue and the > modification is only run once. > (2) Use watchers to do the reporting. > Neither of these are particularly appealing to me as (1) is > asynchronous and (2) appears to slightly complicate things. In my case > it seems a regular old lock would be sufficient? > > There is a discussion on STM generally: > http://groups.google.com/group/clojure/browse_thread/thread/822e21c9a36c5a0f/c81692a8dc1bd198 > But I'm still a little unsure what I should do in this situation. > > > Regards, > Tim. > > > > > --~--~---------~--~----~------------~-------~--~----~ You received this message because you are subscribed to the Google Groups "Clojure" group. To post to this group, send email to clojure@googlegroups.com To unsubscribe from this group, send email to clojure+unsubscr...@googlegroups.com For more options, visit this group at http://groups.google.com/group/clojure?hl=en -~----------~----~----~----~------~----~------~--~---
