Hopbot log for 2008-05-05 - Helma IRC channel: #helma on irc.freenode.net
2008-05-05:
[3:40] <midnightmonster> fyi, Aptana (Eclipse-based web IDE) now has basic support for E4X. That is, it recognizes valid syntax and will flag errors in syntax. Does nothing else useful like auto-insert closing tags or highlight close/open tags yet, but it's a huge start.[14:12] <midnightmonster> I've set loadmode = aggressive on a collection, and the first time I use get(#), it fetches the entire list and uses the cache for subsequent get(#) calls. but every time I call get(name), a new DB query is run to retrieve the child with the matching accessname. I would expect the cache to handle this case as well.
[14:45] <midnightmonster> thing I discovered: you can write whole subqueries in mycollection.filter in type.properties
[14:46] <zumbrunn> yes, I know ;-)
[14:46] <midnightmonster> I didn't know if it would work. I'm glad it does.
[14:46] <zumbrunn> that's why I'm saying type properties docs are a black hole right now
[14:47] <zumbrunn> it's something I'll have to dig into documenting at one point, unless someone else does
[14:49] <zumbrunn> midnightmonster, what's your ichat account name?
[14:49] <zumbrunn> (for tomorrow)
[14:49] <midnightmonster> foolpoet
[14:50] <zumbrunn> ok, still not sure what will happen exactly
[14:50] <midnightmonster> (Still the same AIM name from when I was in high school)
[14:50] <zumbrunn> I don't think you should get up early specially for this
[14:50] <zumbrunn> we can provide you with a bit of a wrap up later
[14:51] <zumbrunn> just tune in when you want
[14:51] <zumbrunn> mine is zumbrunn@mac.com, btw
[14:57] <midnightmonster> cool.
[14:58] <midnightmonster> this is sort of fundamental, but is there a way to create my own object and have it pretend to be an existing HopObject type?
[14:59] <zumbrunn> not sure hwat you mean
[14:59] <zumbrunn> what
[14:59] <midnightmonster> e.g., inherit all that type's methods but have the values that I've already gotten elsewhere
[15:00] <zumbrunn> still not sure what you mean... you mean changing the prototype of an existing object?
[15:00] <midnightmonster> yeah, I guess
[15:01] <midnightmonster> (this would still be interesting to know, but I am an idiot and don't actually need to do this in this case)
[15:02] <zumbrunn> I've played around with this, but don't remember the exact results I got
[15:02] <zumbrunn> I believe I was able to change the objects in memory, but the way they were persisted didn't change
[15:02] <midnightmonster> that's what I want, actually
[15:03] <midnightmonster> (except that I don't really need it as it turns out)
[15:03] <zumbrunn> (I ended up fiddling with the xml files to fix it instead)
[15:08] <midnightmonster> This 200-line report was taking 11 seconds to build. I worked around the fact that get(name) doesn't seem to use the object cache and got it down to 2 seconds, but I can't get it any lower
[15:11] <midnightmonster> and evidently, that's the cost of e4x. if I take out just the tight e4x element generation, it drops to .2 sec
[15:12] <zumbrunn> hmm
[15:17] <midnightmonster> I'm adding 44 TDs one by one to each of 212 TRs. If I build a string with the HTML of the 44 TDs instead and create a new XMLList from the string and append that, I cut the time from 2 sec to 1.4 sec.
[15:17] <midnightmonster> (and I still get output that way, whereas I didn't when I just removed the e4x, obviously)
[15:18] <midnightmonster> someone found, reported, and got at least partly fixed a rhino bug with XMLparsers being created too often instead of reused. do you recall that and know how we could check if that fix is incorporated into the rhino build in helma 1.6.2?
[15:23] <midnightmonster> http://helma.zumbrunn.com/hopbot/2008-01-11
[15:50] <midnightmonster> something I've noticed: for all people's whining over performance issues with closures, looking through calling scopes for variables doesn't seem to be a performance problem at all