Re: Clojure ensure

2010-10-02 Thread ka
I follow this thread, but haven't read Clojure's source. I have a similar question - http://groups.google.com/group/clojure/browse_thread/thread/46415b89c7b311f6, but haven't got any responses yet. -- You received this message because you are subscribed to the Google Groups "Clojure" group. To p

Re: Clojure ensure

2010-09-15 Thread Daniel Gagnon
On Wed, Sep 15, 2010 at 10:44 AM, Stuart Halloway wrote: > Encounter time. I am sitting next to Rich as I write this. > Guaranteed or implementation detail? -- You received this message because you are subscribed to the Google Groups "Clojure" group. To post to this group, send email to clojur

Re: Clojure ensure

2010-09-15 Thread Stuart Halloway
Encounter time. I am sitting next to Rich as I write this. Stu > Ok, I have a 1 and 2 :) > > So is it pessimistic or optimistic? So encounter time ensure or commit > time ensure? > > On Sep 14, 9:36 pm, Stuart Halloway wrote: >> Actually, the second one is basically it. (Except that I don't kn

Re: Clojure ensure

2010-09-14 Thread peter veentjer
I had a quick peek at the stm paper of mark volkmann This function takes a Ref. It prevents other transactions from setting an in-transaction value for the Ref, which also prevents them from committing a new value. This can be used to avoid write skew. It also helps avoid retries due to write con

Re: Clojure ensure

2010-09-14 Thread peter veentjer
Ok, I have a 1 and 2 :) So is it pessimistic or optimistic? So encounter time ensure or commit time ensure? On Sep 14, 9:36 pm, Stuart Halloway wrote: > Actually, the second one is basically it. (Except that I don't know the > Oracle reference, so can't speak to that.) > > Stu > > > The first o

Re: Clojure ensure

2010-09-14 Thread Stuart Halloway
Actually, the second one is basically it. (Except that I don't know the Oracle reference, so can't speak to that.) Stu > The first one is correct. > > On Tue, Sep 14, 2010 at 3:22 PM, peter veentjer wrote: > I have got a question about the Clojure ensure and how it actually > works and the doc

Re: Clojure ensure

2010-09-14 Thread Daniel Gagnon
The first one is correct. On Tue, Sep 14, 2010 at 3:22 PM, peter veentjer wrote: > I have got a question about the Clojure ensure and how it actually > works and the documentation doesn't provide enough information. > > I see a few different solutions: > > 1) An optimistic approach: Once a ref is