Log of the #nice channel on irc.freenode.net

Using timezone: Central European Time
* bonniot leaves01:09
* arjanb leaves01:33
* ArtemGr joins11:31
* bonniot joins16:19
hi Artem16:22
<ArtemGr>salute!16:23
<bonniot>i'm planning to look at the satisfier bug...16:24
<ArtemGr>that's great!
<bonniot>anything else that needs looking into?16:25
<ArtemGr>there is a bug in the type merging, i hadn't time to file it
can send you a stack trace right away if you want16:27
<bonniot>for the type merging? OK
<ArtemGr>but i was going to do a testsuite case from it
<bonniot>testcase would be good yes
it should be a priori easier than the other bug, so it could be a good warming up for me ;-)16:28
the satisfier bug does not happen in the testsuite even in the new framework?16:29
<ArtemGr>didn't tried it in the new framework...
http://rafb.net/paste/results/DcWbuU64.html - type merging stack trace16:32
where do i put the merger case?16:34
typing/inference.testsuite ?
typing/if.testsuite ?
<bonniot>can I see it?16:35
<ArtemGr>i am only going to write it now.16:37
the failing code is here: http://rafb.net/paste/results/ZhPnTl24.html
in the insertionIndex method
<bonniot>a priori, put it in inference16:39
i can move it later once i really know what it's about16:40
or just paste it, and i'll commit it
<ArtemGr>okay, i'll do the latter16:41
type merging test case: http://rafb.net/paste/results/OVDShv76.html17:03
<bonniot>ok, reproduced17:05
... and fixed ;-)17:09
<ArtemGr>i *knew* it's an easy one. : )17:10
<bonniot>merging with Object was just not handled properly17:21
i wrote a second testcase that exercises a cousin situation (merging Object with non-Object)
the fix: http://rafb.net/paste/results/e1fjj478.html17:24
<ArtemGr>as you see, i'll try to write a faster HashMap, somewhat like http://trove4j.sourceforge.net/17:25
<bonniot>now let's see how my fanless laptop handles a bootstrap ;-)
cool17:26
are you also implementing the primitive type versions?
<ArtemGr>it is a special model : ) , or the fan just died?17:27
don't know yet. i have a lot of integer keyed maps, but first target is to made an Object one and then a TimeCache on top of it, with ARC caching strategy (instead of LRU).17:28
<bonniot>actually, I unplugged the fan ;-)17:29
it's an experiment
keeping it cool with lowering the freq and voltage when necessary
hum, there is a bug...17:31
<ArtemGr>is it under Windows? a friend of mine is using speedfan for this, although he tried a bit of other programs before.17:32
<bonniot>linux
<ArtemGr>cool17:33
<bonniot>as far as I know, this motherboard does not support software control of the fan, unfortunately
<ArtemGr>speedfan got frequency control recently..
<bonniot>but the proc support control of clock and voltage
<ArtemGr>ah, that's nice of linux17:34
<bonniot>and I modified the control software to take temperature into account
<ArtemGr>perhaps i'll try to do this with speedfan on my laptop : ) (although without unplugging the fan)17:39
<bonniot>if you can control it, it's definitely better than unpluggin ;-)17:43
<ArtemGr>i'm not sure about it, but perhaps i can reduce the fan usage by lowering frequency before the fan gets into play17:46
<bonniot>yes, that's a possible solution17:49
you can do that with my patch of cpudyn for linux, by setting the right temperature
(my fan kicks at 53C)
<ArtemGr>thanks, but i have XP currently and going to use FreeBSD in the future.17:50
<bonniot>should work with BSD17:51
ah, depending on where the info is, might need some adaptation17:52
<ArtemGr>yeah, i don't think it would work out of the box with FreeBSD, hardware interfaces are totally different here17:53
<bonniot>what processor do you have?17:55
<ArtemGr>Intel-M, but FreeBSD i'm going to use a bit later, preferentially when dual-processed laptops become common. that's my idea fix, to have a dual-processed FreeBSD laptop to debug multithreaded Java apps. 8 )17:57
<bonniot>pentium-m or celeron-m ?17:58
<ArtemGr>Intel Pentium III-M Tualatin
<bonniot>ah, I think I found the bug18:18
<ArtemGr>the one in the type merging?18:25
<bonniot>yes
the first fix was slightly incorrect
* ArtemGr leaves18:35
<CIA-2>03bonniot * 10Nice/ (2 files in 2 dirs): Fix mergings involving the Object type.21:15
03bonniot * 10Nice/regtest/regtest: 21:58
Use en_US instead of french locale, since that's more "standard"
(anything with latin1 characters is OK).
* bonniot leaves23:59