Hm. You say you are following lj_dev, but you seem to be missing
something. First, I am not running any servers, let alone LiveJournal ones, apart from my little development installation which I use only for testing. Hopefully, I'll have a Wikipedia one soon too. Secondly, both MemCacheD and FotoBilder (and S2 as well) were programmed almost entirely by Bradfitz and Avva; I and other volunteers were not invited to participate in the development of it. .<<<<
whoops, i see your posts so often in lj-dev, and your name so much in bugzilla, and since you've had a higher ranking in the bazaar than avva, i kinda assumed you had become part of the core team, and maybe worked for LJ. my bad.
Anyways.. do you see a possibility of memcacheD ever being useful in the wikipedia? obviously at this point it can't really be used because we would need a ton more memory and probably another server, but, would it be practical for the large ammount of text per article, and large ammount of constantly changing articles that wikipedia has? from what I assume, given that there was enough memory available it would do wonders for the article search, woulnt it?
________________________________________________ This mail was sent by UebiMiau 2.5
Hi,
i see your posts so often in lj-dev
You could say I'm trying to force my popularity, LOL ;-)
and your name so much in bugzilla
That's because I enjoy fixing tiny trivial little buggies, er, buglets, or whatever the diminutive of bug is. Naturally, I can fix more of those in the same time as I can write entire features (which I also enjoy doing, but they're not very welcome on LJ).
, and since you've had a higher ranking in the bazaar than avva, i kinda assumed you had become part of the core team, and maybe worked for LJ. my bad.
Heheh. The Bazaar doesn't make someone a core member. :) I've been almost hermetically kept away from the core team, you'd think they have an aversion against me :-)
Anyways.. do you see a possibility of memcacheD ever being useful in the wikipedia?
Someone would have to translate the API to PHP.
obviously at this point it can't really be used because we would need a ton more memory and probably another server, but, would it be practical for the large ammount of text per article, and large ammount of constantly changing articles that wikipedia has?
Average articles on Wikipedia aren't that much larger than average LiveJournal entries. Although you're quite right in saying that Wikipedia articles are updated quite a lot more often than LiveJournal entries, this is balanced by the fact that new LiveJournal entries are produced in much greater quantities than Wikipedia articles. I seem to remember having read somewhere that all of Wikipedia's articles in their most recent versions total 4 GB or something? Contrast that with LiveJournal's 80 GB ;-). Readers would benefit from MemCacheD no matter what, even if editors don't (although I'm sure editors will, too).
from what I assume, given that there was enough memory available it would do wonders for the article search, woulnt it?
Actually, this is an interesting point. I'm not sure MemCacheD will help here because it certainly does not build an in-memory keyword search index. It's just a cache, not an in-memory database :-). I don't know how the search function works at Wikipedia (LiveJournal doesn't have one), but the way I understand it, it performs an SQL query (disregarding the Go button for now) that cannot be "translated" to MemCacheD.
So, maybe the search will still have to be done on the database. But I'd be very glad if we could speed up the servers by MemCacheD'ing everything else.
Greetings, Timwi
wikitech-l@lists.wikimedia.org