2014-12-07T00:04:06Z attila_lendvai quit (Quit: Leaving.)
2014-12-07T00:21:12Z echo-area joined #sbcl
2014-12-07T00:37:33Z eudoxia joined #sbcl
2014-12-07T00:38:58Z eudoxia quit (Client Quit)
2014-12-07T00:50:51Z LiamH quit (Quit: Leaving.)
2014-12-07T01:02:40Z DeadTrickster quit (Ping timeout: 255 seconds)
2014-12-07T01:19:43Z pkhuong: committed.
2014-12-07T02:46:06Z edgar-rft joined #sbcl
2014-12-07T03:33:18Z |3b|: i guess https://bugs.launchpad.net/sbcl/+bug/1393740 applies to linux too
2014-12-07T03:34:27Z |3b|: and fix there seems to work as well
2014-12-07T03:39:34Z christoph_debian quit (Ping timeout: 258 seconds)
2014-12-07T03:41:14Z psy_ quit (Ping timeout: 244 seconds)
2014-12-07T03:43:23Z echo-area quit (Remote host closed the connection)
2014-12-07T03:44:04Z echo-area joined #sbcl
2014-12-07T03:53:31Z christoph_debian joined #sbcl
2014-12-07T05:20:17Z leo2007 quit (Ping timeout: 240 seconds)
2014-12-07T05:26:09Z psy_ joined #sbcl
2014-12-07T05:29:10Z momo-reina joined #sbcl
2014-12-07T05:35:36Z momo-reina quit (Remote host closed the connection)
2014-12-07T05:39:55Z pppp2 joined #sbcl
2014-12-07T05:42:30Z karswell` joined #sbcl
2014-12-07T05:45:29Z karswell quit (Ping timeout: 264 seconds)
2014-12-07T06:11:01Z DeadTrickster joined #sbcl
2014-12-07T06:18:59Z scymtym_ quit (Ping timeout: 245 seconds)
2014-12-07T06:57:51Z pppp2 quit (Read error: No route to host)
2014-12-07T07:04:54Z stassats` joined #sbcl
2014-12-07T07:20:38Z momo-reina joined #sbcl
2014-12-07T07:35:43Z psy_ quit (Ping timeout: 252 seconds)
2014-12-07T07:39:55Z drmeister quit (Ping timeout: 258 seconds)
2014-12-07T07:41:06Z leo2007 joined #sbcl
2014-12-07T08:36:12Z pppp2 joined #sbcl
2014-12-07T08:38:24Z ASau` joined #sbcl
2014-12-07T08:41:56Z ASau quit (Ping timeout: 244 seconds)
2014-12-07T09:29:01Z momo-reina quit (Ping timeout: 250 seconds)
2014-12-07T09:29:19Z psy_ joined #sbcl
2014-12-07T09:35:39Z milosn quit (Ping timeout: 264 seconds)
2014-12-07T09:51:20Z echo-area quit (Remote host closed the connection)
2014-12-07T10:21:53Z stassats` quit (Ping timeout: 250 seconds)
2014-12-07T10:27:48Z leo2007 quit (Remote host closed the connection)
2014-12-07T10:38:47Z oleo is now known as Guest6095
2014-12-07T10:39:35Z oleo__ joined #sbcl
2014-12-07T10:41:37Z Guest6095 quit (Ping timeout: 240 seconds)
2014-12-07T10:46:50Z milosn joined #sbcl
2014-12-07T11:06:52Z leo2007 joined #sbcl
2014-12-07T11:23:56Z eudoxia joined #sbcl
2014-12-07T11:37:13Z angavrilov joined #sbcl
2014-12-07T11:54:43Z eudoxia quit (Quit: leaving)
2014-12-07T12:13:04Z ASau` is now known as ASau
2014-12-07T12:13:59Z drmeister joined #sbcl
2014-12-07T12:39:17Z drmeister quit (Ping timeout: 272 seconds)
2014-12-07T12:47:47Z attila_lendvai joined #sbcl
2014-12-07T13:09:20Z pppp2 quit (Read error: Connection reset by peer)
2014-12-07T13:22:05Z oleo__ quit (Quit: Verlassend)
2014-12-07T13:28:11Z attila_lendvai quit (Quit: Leaving.)
2014-12-07T14:32:27Z drmeister joined #sbcl
2014-12-07T14:37:05Z stassats` joined #sbcl
2014-12-07T14:49:05Z milosn quit (Ping timeout: 264 seconds)
2014-12-07T14:53:03Z oleo joined #sbcl
2014-12-07T14:54:31Z pkhuong: stassats`: re 1398785
2014-12-07T14:55:34Z pkhuong: what do you think of changing the behaviour depending on the speed/space quality?
2014-12-07T14:56:09Z pkhuong: currently, representation selection sums costs for every VOP
2014-12-07T14:56:24Z stassats`: i have no idea
2014-12-07T14:56:36Z stassats`: i only can see the best result, but not how to achieve it
2014-12-07T14:56:50Z pkhuong: that makes sense if we're optimising for space, or multiple VOPs write to the same variable in sequence
2014-12-07T14:57:01Z pkhuong: (or read from the same variable)
2014-12-07T14:57:13Z pkhuong: most of the time, I see mulitple writes and reads because of branches
2014-12-07T14:57:43Z pkhuong: so I can instead add the average cost for writes and the average for read vops
2014-12-07T14:57:47Z pkhuong: and minimise that
2014-12-07T14:57:58Z stassats`: can we somehow determine shared paths to a destination and unite them?
2014-12-07T14:58:30Z pkhuong: not easily
2014-12-07T14:59:29Z stassats`: is the representation and vop selection happening at too late a stage for that?
2014-12-07T14:59:55Z pkhuong: it's more that we don't have that kind of pass
2014-12-07T15:00:23Z stassats`: vop and representation selection is what always bothered me
2014-12-07T15:00:41Z pkhuong: they're really bad greedy heuristics
2014-12-07T15:13:29Z stassats`: could we have movable stacks on the precise gengc?
2014-12-07T15:15:16Z pkhuong: doubt that. there's still DX allocation
2014-12-07T15:15:39Z stassats`: but the pointers could be fixed up
2014-12-07T15:15:48Z stassats`: use case: growing stacks
2014-12-07T15:16:03Z stassats`: dynamically upon exhaustion
2014-12-07T15:16:44Z pkhuong: seems simpler to have segmented stacks
2014-12-07T15:17:08Z stassats`: not that ever head problems with the stack
2014-12-07T15:17:11Z stassats`: i
2014-12-07T15:17:15Z stassats`: had, sigh
2014-12-07T15:17:30Z pkhuong: I think rust used to segment and decided to just go for a normal stack
2014-12-07T15:17:44Z pkhuong: there might be some good discussions of the pros and cons somewhere
2014-12-07T15:17:54Z pkhuong: Not sure what go's doing
2014-12-07T15:17:55Z stassats`: maybe it can calculate the minimum size?
2014-12-07T15:21:12Z Krystof: I'd like a growable heap first :-(
2014-12-07T15:21:57Z stassats`: relocatable is more useful
2014-12-07T15:35:16Z krzysz00 quit (Quit: Lost terminal)
2014-12-07T15:35:46Z pkhuong: stassats`: can you apply http://paste.lisp.org/display/144622 and see if you think it improves more things than it worsens?
2014-12-07T15:39:44Z pkhuong: (at speed > space, and speed > 1)
2014-12-07T15:54:23Z leo2007 quit (Quit: rcirc on GNU Emacs 25.0.50.1)
2014-12-07T16:08:52Z edgar-rft quit (Quit: the consequences are unspecified)
2014-12-07T16:37:17Z ASau quit (Remote host closed the connection)
2014-12-07T16:38:42Z ASau joined #sbcl
2014-12-07T17:26:42Z scymtym_ joined #sbcl
2014-12-07T17:33:45Z stassats`: does it affect anything else?
2014-12-07T17:45:29Z karswell` quit (Ping timeout: 252 seconds)
2014-12-07T17:48:42Z karswell` joined #sbcl
2014-12-07T17:57:36Z attila_lendvai joined #sbcl
2014-12-07T18:28:00Z karswell` is now known as karswell
2014-12-07T18:32:53Z milosn joined #sbcl
2014-12-07T18:33:12Z njmurphy joined #sbcl
2014-12-07T19:25:25Z milosn quit (Read error: Connection reset by peer)
2014-12-07T19:25:56Z milosn joined #sbcl
2014-12-07T19:31:50Z attila_lendvai quit (Quit: Leaving.)
2014-12-07T19:35:34Z LiamH joined #sbcl
2014-12-07T19:52:48Z attila_lendvai joined #sbcl
2014-12-07T19:55:28Z stassats` quit (Ping timeout: 264 seconds)
2014-12-07T20:05:56Z ASau quit (Remote host closed the connection)
2014-12-07T20:07:06Z ASau joined #sbcl
2014-12-07T20:15:03Z karswell quit (Ping timeout: 265 seconds)
2014-12-07T20:20:53Z pkhuong: all the code with more than 1 write or read on a tn (w/ the appropriate policy)
2014-12-07T20:43:23Z pppp2 joined #sbcl
2014-12-07T20:57:42Z stassats` joined #sbcl
2014-12-07T21:04:09Z KingCons_ joined #sbcl
2014-12-07T21:04:27Z redline6561 quit (Quit: ZNC - http://znc.in)
2014-12-07T21:04:52Z KingCons_ quit (Quit: ZNC - http://znc.in)
2014-12-07T21:05:36Z redline6561 joined #sbcl
2014-12-07T21:11:23Z pppp2 quit (Read error: Connection reset by peer)
2014-12-07T21:19:49Z milosn quit (Ping timeout: 245 seconds)
2014-12-07T21:30:20Z stassats` quit (Quit: ERC Version 5.3 (IRC client for Emacs))
2014-12-07T21:39:22Z angavrilov quit (Remote host closed the connection)
2014-12-07T21:40:14Z milosn joined #sbcl
2014-12-07T21:44:26Z ASau quit (Remote host closed the connection)
2014-12-07T21:47:46Z ASau joined #sbcl
2014-12-07T22:14:45Z ASau quit (Remote host closed the connection)
2014-12-07T22:21:11Z ASau joined #sbcl
2014-12-07T22:54:19Z oleo is now known as Guest32365
2014-12-07T22:55:11Z oleo__ joined #sbcl
2014-12-07T22:56:14Z karswell joined #sbcl
2014-12-07T22:57:14Z Guest32365 quit (Ping timeout: 244 seconds)
2014-12-07T22:58:37Z pillton: Re 1398785: http://hastebin.com/alilodupoh
2014-12-07T22:59:35Z pillton: I think that optimization is wrong.
2014-12-07T23:04:55Z jsnell: looks ok to me
2014-12-07T23:05:27Z jsnell: eq on two numbers can always return nil, no matter what
2014-12-07T23:06:30Z pillton: Both functions return the argument so they should be EQ.
2014-12-07T23:06:50Z pillton: Oh wait. "An implementation is permitted to make ``copies'' of characters and numbers at any time. "
2014-12-07T23:06:54Z pillton: Sorry.
2014-12-07T23:07:36Z pillton: That is weird.
2014-12-07T23:12:45Z pillton: I wonder what the history of that sentence is. That means EQ can return NIL if both of its arguments are integers.
2014-12-07T23:12:59Z pillton: s/integers/numbers/
2014-12-07T23:13:48Z |3b|: right, are 2 numbers in different registers the 'same' integer in the sense that 2 EQ conses are the 'same' cons?
2014-12-07T23:14:25Z |3b|: or 2 bignums in different addresses?
2014-12-07T23:15:33Z pillton: I see what you are saying, but consider (eq v (identity v)).
2014-12-07T23:15:51Z |3b|: EQ probably compares by address, so 2 copies of same value bignum would need handled specially, and long ago that was probably expensive
2014-12-07T23:15:54Z jsnell: it simply means that EQ is not a very meaningful operation with value types, and it happens that in CL the characters and numbers are the only value types in CL
2014-12-07T23:15:55Z pillton: That is NIL for all numbers.
2014-12-07T23:16:19Z |3b|: you also need to deal with comparing boxed vs unboxed fixnums
2014-12-07T23:16:47Z jsnell: yes, if you're comparing numbers or characters, and you're not interested in object identity but value equality for those data types, you should use EQL
2014-12-07T23:17:02Z jsnell: well, actually you should always use EQL rather than EQ unless you understand what EQ does
2014-12-07T23:17:18Z pillton: I know the difference.
2014-12-07T23:17:23Z |3b|: yeah, never using EQ is probably reasonably advice these days
2014-12-07T23:17:30Z jsnell: you clearly did not know it 20 minutes ago :-)
2014-12-07T23:18:06Z pillton: I was unaware of the note.
2014-12-07T23:20:19Z pillton: It just seems weird to have a thing that is not EQ to itself.
2014-12-07T23:20:46Z |3b|: it isn't... it is a thing for which EQ is undefined
2014-12-07T23:20:48Z jsnell: clearly it's not "itself", it's another object
2014-12-07T23:21:18Z |3b|: which is the entire difference between EQ and EQL, EQ is undefined on some things
2014-12-07T23:22:25Z attila_lendvai quit (Quit: Leaving.)
2014-12-07T23:30:15Z jsnell: clearly it's not "itself", it's another object
2014-12-07T23:30:23Z jsnell: oops
2014-12-07T23:41:15Z pillton: I really don't like it. http://hastebin.com/zufulewoce
2014-12-07T23:41:23Z pillton: Anyway, I'll move on.
2014-12-07T23:41:26Z pillton: Thanks.