Memcache CAS

Chris Goffinet goffinet at yahoo-inc.com
Mon Sep 17 07:44:54 UTC 2007


Here is the fix Dustin, I can't believe I missed the code to drain the  
data line, it was right below :-) Here is the new diff. Its a clean  
diff from original code to this, and it won't make you maintain 2  
versions :)

I will look into the global counter / memory address issue tomorrow  
and get out a fix.

-------------- next part --------------
A non-text attachment was scrubbed...
Name: memcache.cas.drain.diff
Type: application/octet-stream
Size: 5953 bytes
Desc: not available
Url : http://lists.danga.com/pipermail/memcached/attachments/20070917/aa3a7d2e/memcache.cas.drain.obj
-------------- next part --------------




Chris Goffinet
goffinet at yahoo-inc.com



On Sep 17, 2007, at 12:33 AM, Dustin Sallings wrote:

>
> On Sep 17, 2007, at 0:31, Chris Goffinet wrote:
>
>> 1) Okay. Does it make sense then to just implement a 'revision'  
>> into the item struct? That way we can just revision++ on new 'set'?
>
> 	I think you'd need a global counter.  You wouldn't want a set ->  
> delete -> set to tell you that your value is the same between those  
> two sets.
>
>> 2) I'll add support for those 3 again so you can see what happened.  
>> I'll look into draining the output correctly so client's won't have  
>> an issue with this.
>
> 	There's a drain state, but it ends strangely for the text protocol,  
> I believe.
>
> -- 
> Dustin Sallings
>
>



More information about the memcached mailing list