Excellent! Waiting to test it out. Currently I am using 1.2.4. I would be testing it only using the text protocol since binary one is yet to be released.
-Rakesh On Fri, Feb 22, 2008 at 3:47 PM, Dustin Sallings <[EMAIL PROTECTED]> wrote: > > I did a first pass at a client implementation in a branch at the airport > this morning or last night or whatever you call that. When I get near a > network I'll try to make it available to you to look at. > > The only overhead is that I have to wrap the results in another object > to give you both parts. The call in the binary protocol is the same (same > for the actual CAS op... I didn't have to write much to get it working in > bin). In the text protocol, it's still not worth worrying about. > > I didn't do any type of bulk gets on the client interface, but the > aggregation part of the request optimization could do it for you. One > optimization that I intend to make here is to turn a series of mixed get and > gets requests into a single bulk gets on the text protocol. This is > unnecessary on the bin protocol because it's already the same on the wire, > so it just figures out whether you want the CAS ID or not at the time of > receiving a result. > > -- Dustin Sallings (mobile) > > On Feb 22, 2008, at 2:10, "Rakesh Rajan" <[EMAIL PROTECTED]> wrote: > > Thanks for the reply! Are there any performance implication of using a > normal get v/s a get that also returns the unique cas number ? > > Looking forward to CAS implementation in the client. Until then I will > stick to the current locking method for which I would love to hear if there > are any improvements that I could make. > > -Rakesh > > On Fri, Feb 22, 2008 at 3:31 AM, Dustin Sallings < <[EMAIL PROTECTED]> > [EMAIL PROTECTED]> wrote: > > > > > On Feb 21, 2008, at 12:51, Rakesh Rajan wrote: > > > > > I am currently using Dustin's Java library (memcached-2.0-pre7). I > > > havn't seen any CAS function in that. So safely assuming that I need > > > to continue to use java for my project ( :) ) , what are the > > > alternatives that I have ? > > > > > > So from my understanding of CAS, I can pass a unique code, that I > > > get from "get" operation, to the "set" operation. And in case it > > > fails, I would need to fetch the latest value and update the cache > > > entry again ( this would need to some kind of loop to make sure that > > > I don't miss pushing any updates to cache). > > > > > > Sorry, I've been a bit behind in my emails. > > > > No, I never quite worked CAS support into my java client. I've > > got a > > long plane ride ahead of me tomorrow, so maybe that'll give me > > something to do. :) > > > > Basically, you do the get+set in a loop where the set fails if > > something else mutated the value between the get and set. > > > > -- > > Dustin Sallings > > > > >
