<html><body style="word-wrap: break-word; -webkit-nbsp-mode: space; -webkit-line-break: after-white-space; ">
Dustin,<div><br class="webkit-block-placeholder"></div><div>I have the prototype of the CAS operation working from the latest memcache trunk in C using text protocol. I have also patched libmemcache to do support "gets" and "cas". I will be doing a cleanup, I can send you the patches later tonight.</div><div><br><div contenteditable="false"> <span class="Apple-style-span" style="border-collapse: separate; border-spacing: 0px 0px; color: rgb(0, 0, 0); font-family: Helvetica; font-size: 12px; font-style: normal; font-variant: normal; font-weight: normal; letter-spacing: normal; line-height: normal; text-align: auto; -khtml-text-decorations-in-effect: none; text-indent: 0px; -apple-text-size-adjust: auto; text-transform: none; orphans: 2; white-space: normal; widows: 2; word-spacing: 0px; "><br class="Apple-interchange-newline"><div><br class="khtml-block-placeholder"></div><div>Chris Goffinet</div><div><a href="mailto:goffinet@yahoo-inc.com">goffinet@yahoo-inc.com</a></div><div><br class="khtml-block-placeholder"></div><br class="Apple-interchange-newline"></span> </div><br><div><div>On Sep 8, 2007, at 11:26 AM, Dustin Sallings wrote:</div><br class="Apple-interchange-newline"><blockquote type="cite"><div style="word-wrap: break-word; -khtml-nbsp-mode: space; -khtml-line-break: after-white-space; "><div><br class="khtml-block-placeholder"></div><div>[replying back on the list to try to keep the discussion open and hopefully avoid repeating myself]</div><br><div><div>On Sep 8, 2007, at 7:31, Jehiah Czebotar wrote:</div><br class="Apple-interchange-newline"><blockquote type="cite"><div style="margin-top: 0px; margin-right: 0px; margin-bottom: 0px; margin-left: 0px; "><font face="Helvetica" size="3" style="font: 12.0px Helvetica">This sounds like the proposed "append" method which would allow</font></div><div style="margin-top: 0px; margin-right: 0px; margin-bottom: 0px; margin-left: 0px; "><font face="Helvetica" size="3" style="font: 12.0px Helvetica">memcache to handle lists on the server side... with one caviat, you</font></div><div style="margin-top: 0px; margin-right: 0px; margin-bottom: 0px; margin-left: 0px; "><font face="Helvetica" size="3" style="font: 12.0px Helvetica">want a conditional append which only appends if the value isn't</font></div><div style="margin-top: 0px; margin-right: 0px; margin-bottom: 0px; margin-left: 0px; "><font face="Helvetica" size="3" style="font: 12.0px Helvetica">already there.</font></div></blockquote><div><br class="khtml-block-placeholder"></div><div><span class="Apple-tab-span" style="white-space:pre"> </span>In this example usage...sort of. One would also want to remove items from the list to keep it from growing out of control.</div><div><br class="khtml-block-placeholder"></div><div><span class="Apple-tab-span" style="white-space:pre"> </span>There was another proposal to allow updating parts of a value by offset or other such things. None of these things can work race-free without implementing advisory locks which, as already pointed out, may be subject to the LRU.</div><br><blockquote type="cite"><div style="margin-top: 0px; margin-right: 0px; margin-bottom: 0px; margin-left: 0px; "><font face="Helvetica" size="3" style="font: 12.0px Helvetica">If memcached were handling the lists, you wouldn't have to add all</font></div><div style="margin-top: 0px; margin-right: 0px; margin-bottom: 0px; margin-left: 0px; "><font face="Helvetica" size="3" style="font: 12.0px Helvetica">that conditional logic and "replace if it's still the same" stuff</font></div><div style="margin-top: 0px; margin-right: 0px; margin-bottom: 0px; margin-left: 0px; "><font face="Helvetica" size="3" style="font: 12.0px Helvetica">which won't work in high concurrency situations anyway.</font></div> </blockquote></div><div><br class="khtml-block-placeholder"></div><div><span class="Apple-tab-span" style="white-space:pre"> </span>Well, it's not entirely fair to say it doesn't work in high concurrency situations. The operation is always atomic, but it may need to be retried. This is a common strategy in concurrent programming for creating lock-free concurrent data structures. If a particular key is too hot, though, you could potentially split it into multiple keys to reduce retries in practice. In the current example, you could have a series of current_login keys hashed by uid. Your contention is reduced by the number of keys. There's never a case where you blindly overwrite values, though.</div><br><div><span class="Apple-tab-span" style="white-space:pre"> </span>Adding an add-this-to-a-list-if-it-does-not-exist-and-remove-something-else-to-keep-the-list-to-n-elements would be an unreasonable operation to add. The only reasonable way to do something like this would be to add server-side scripting and then effectively write the exact operations you need so that they may be performed atomically.</div><div><br class="khtml-block-placeholder"></div><div><span class="Apple-tab-span" style="white-space:pre"> </span>Are there enough potential server-side operations to really justify a scripting interface? When it's been brought up in meetings, the response has varied from interest, to a warning about crossing streams or something.</div><div><br class="khtml-block-placeholder"></div><div><span class="Apple-tab-span" style="white-space:pre"> </span>Both a CAS and a scripting interface provide a foundation for building atomic data structure manipulation without having to define all of the data structures and all of the operations at the protocol level. Each has downside. With CAS, you may have to retry a few times under contention and this would be amplified by the latency. With scripting, you may easily slow down the whole server, and you have to deal with how parameters are passed in and results are returned -- and this would mean very different things by protocol.</div><div><br class="khtml-block-placeholder"></div><div>-- </div><div>Dustin</div></div></blockquote></div><br></div></body></html>