2015-01-22T00:07:47Z ehaliewicz joined #sbcl
2015-01-22T00:44:41Z slyrus quit (Ping timeout: 246 seconds)
2015-01-22T01:00:33Z karswell` joined #sbcl
2015-01-22T01:00:55Z karswell quit (Remote host closed the connection)
2015-01-22T01:06:57Z Bicyclidine quit (Ping timeout: 245 seconds)
2015-01-22T01:08:14Z edgar-rft quit (Quit: continuation abandoned because no happening found)
2015-01-22T01:09:17Z attila_lendvai quit (Quit: Leaving.)
2015-01-22T01:36:18Z nikki93 joined #sbcl
2015-01-22T01:43:58Z nikki93 quit (Remote host closed the connection)
2015-01-22T01:56:27Z weissschloss quit (Ping timeout: 264 seconds)
2015-01-22T01:57:13Z Bike_ joined #sbcl
2015-01-22T01:59:31Z pkhuong_ joined #sbcl
2015-01-22T02:03:16Z loke: pkhuong: Thanks for your reply. It took until now for me to read it, but please know that I did and I appreciate it.
2015-01-22T02:06:02Z Bike quit (Ping timeout: 264 seconds)
2015-01-22T02:06:03Z pkhuong quit (Ping timeout: 264 seconds)
2015-01-22T02:08:06Z weissschloss joined #sbcl
2015-01-22T02:51:43Z hlavaty quit (Remote host closed the connection)
2015-01-22T03:03:00Z ehaliewicz quit (Ping timeout: 265 seconds)
2015-01-22T03:25:27Z Bike_ is now known as Bike
2015-01-22T03:38:29Z christop1 quit (Ping timeout: 256 seconds)
2015-01-22T03:40:43Z dekhoda joined #sbcl
2015-01-22T03:51:50Z christop1 joined #sbcl
2015-01-22T04:00:11Z psy_ quit (Quit: Leaving)
2015-01-22T04:03:12Z psy_ joined #sbcl
2015-01-22T04:55:56Z nyef quit (Quit: G'night all)
2015-01-22T05:02:28Z foom quit (Ping timeout: 245 seconds)
2015-01-22T05:14:51Z foom joined #sbcl
2015-01-22T05:20:43Z echo-area quit (Remote host closed the connection)
2015-01-22T05:25:13Z echo-area joined #sbcl
2015-01-22T05:31:23Z gingerale quit (Ping timeout: 265 seconds)
2015-01-22T05:43:55Z gingerale joined #sbcl
2015-01-22T06:30:49Z sdemarre joined #sbcl
2015-01-22T06:35:49Z psy_ quit (Quit: Leaving)
2015-01-22T06:50:00Z oleo quit (Quit: Verlassend)
2015-01-22T06:58:50Z angavrilov joined #sbcl
2015-01-22T07:00:01Z slyrus joined #sbcl
2015-01-22T07:08:55Z Shinmera joined #sbcl
2015-01-22T07:20:55Z sdemarre quit (Ping timeout: 264 seconds)
2015-01-22T07:54:51Z chu quit (Ping timeout: 276 seconds)
2015-01-22T07:55:53Z chu joined #sbcl
2015-01-22T08:04:32Z psy_ joined #sbcl
2015-01-22T08:05:09Z psy_ quit (Max SendQ exceeded)
2015-01-22T08:05:50Z psy_ joined #sbcl
2015-01-22T08:51:58Z cojy quit (Ping timeout: 265 seconds)
2015-01-22T08:56:03Z cojy joined #sbcl
2015-01-22T09:05:31Z Posterdati quit (Ping timeout: 255 seconds)
2015-01-22T09:18:59Z Posterdati joined #sbcl
2015-01-22T09:35:11Z karswell` quit (Remote host closed the connection)
2015-01-22T09:44:13Z chu quit (Ping timeout: 255 seconds)
2015-01-22T09:46:00Z chu joined #sbcl
2015-01-22T10:08:32Z pacon joined #sbcl
2015-01-22T10:14:06Z stassats joined #sbcl
2015-01-22T10:28:59Z hlavaty joined #sbcl
2015-01-22T11:59:44Z chu quit (Remote host closed the connection)
2015-01-22T12:00:00Z chu joined #sbcl
2015-01-22T12:19:13Z eudoxia joined #sbcl
2015-01-22T12:53:34Z pacon quit (Read error: Connection reset by peer)
2015-01-22T13:10:41Z edgar-rft joined #sbcl
2015-01-22T13:19:55Z psy_ quit (Quit: Leaving)
2015-01-22T13:24:57Z Quadrescence quit (Quit: This computer has gone to sleep)
2015-01-22T13:34:58Z Adlai joined #sbcl
2015-01-22T13:56:35Z Adlai quit (Ping timeout: 250 seconds)
2015-01-22T13:57:43Z Adlai joined #sbcl
2015-01-22T13:57:58Z Adlai quit (Client Quit)
2015-01-22T14:06:52Z psilord1 quit (Quit: Leaving.)
2015-01-22T14:25:56Z pacon joined #sbcl
2015-01-22T14:40:40Z pacon quit (Read error: Connection reset by peer)
2015-01-22T14:41:29Z oleo joined #sbcl
2015-01-22T15:01:36Z psy joined #sbcl
2015-01-22T15:45:06Z nyef joined #sbcl
2015-01-22T15:51:24Z karswell` joined #sbcl
2015-01-22T16:14:12Z edgar-rft quit (Quit: computation stopped into eternal vacuum)
2015-01-22T16:14:31Z chu quit (Ping timeout: 244 seconds)
2015-01-22T16:20:31Z chu joined #sbcl
2015-01-22T16:22:07Z attila_lendvai joined #sbcl
2015-01-22T16:56:57Z chu quit (Ping timeout: 276 seconds)
2015-01-22T16:57:31Z chu joined #sbcl
2015-01-22T16:59:20Z edgar-rft joined #sbcl
2015-01-22T17:00:30Z eudoxia quit (Quit: Leaving)
2015-01-22T17:07:46Z chu quit (Read error: Connection reset by peer)
2015-01-22T17:13:03Z Bicyclidine joined #sbcl
2015-01-22T17:14:10Z Adlai joined #sbcl
2015-01-22T17:17:13Z edgar-rft quit (Quit: execution vanished because everything exploded)
2015-01-22T17:21:14Z Adlai quit (Remote host closed the connection)
2015-01-22T17:21:23Z Adlai` joined #sbcl
2015-01-22T17:25:04Z Adlai` is now known as adlai
2015-01-22T17:39:24Z karswell` quit (Read error: Connection reset by peer)
2015-01-22T17:39:39Z karswell` joined #sbcl
2015-01-22T18:07:39Z adlai quit (Remote host closed the connection)
2015-01-22T18:09:24Z adlai joined #sbcl
2015-01-22T18:14:34Z sdemarre joined #sbcl
2015-01-22T18:18:29Z edgar-rft joined #sbcl
2015-01-22T18:22:27Z karswell` quit (Read error: Connection reset by peer)
2015-01-22T18:22:45Z karswell` joined #sbcl
2015-01-22T18:24:23Z psy quit (Disconnected by services)
2015-01-22T18:24:43Z psy_ joined #sbcl
2015-01-22T18:51:05Z slyrus quit (Ping timeout: 246 seconds)
2015-01-22T18:52:28Z karswell` quit (Ping timeout: 245 seconds)
2015-01-22T18:54:06Z adlai quit (Remote host closed the connection)
2015-01-22T18:54:25Z Adlai joined #sbcl
2015-01-22T19:07:28Z gabriel_laddel quit (Ping timeout: 245 seconds)
2015-01-22T19:07:42Z nikki93 joined #sbcl
2015-01-22T19:26:39Z nikki93 quit (Remote host closed the connection)
2015-01-22T19:27:12Z nikki93 joined #sbcl
2015-01-22T19:28:51Z sdemarre quit (Ping timeout: 264 seconds)
2015-01-22T19:32:18Z nikki93 quit (Ping timeout: 276 seconds)
2015-01-22T19:32:18Z psilord joined #sbcl
2015-01-22T19:35:11Z nikki93 joined #sbcl
2015-01-22T19:39:27Z stassats quit (Quit: ERC Version 5.3 (IRC client for Emacs))
2015-01-22T19:40:18Z Adlai quit (Remote host closed the connection)
2015-01-22T19:41:34Z Adlai joined #sbcl
2015-01-22T19:42:35Z nikki93 quit (Remote host closed the connection)
2015-01-22T19:56:11Z Bicyclidine quit (Ping timeout: 252 seconds)
2015-01-22T20:02:55Z Bicyclidine joined #sbcl
2015-01-22T20:10:14Z Bicyclidine quit (Ping timeout: 245 seconds)
2015-01-22T20:18:32Z Bicyclidine joined #sbcl
2015-01-22T20:21:18Z nikki93 joined #sbcl
2015-01-22T20:28:19Z Adlai quit (Ping timeout: 250 seconds)
2015-01-22T20:33:05Z Adlai joined #sbcl
2015-01-22T20:33:11Z nikki93 quit (Remote host closed the connection)
2015-01-22T20:33:45Z nikki93 joined #sbcl
2015-01-22T20:36:01Z nikki93_ joined #sbcl
2015-01-22T20:36:54Z nikki93 quit (Read error: Connection reset by peer)
2015-01-22T20:43:32Z nikki93_ quit (Remote host closed the connection)
2015-01-22T20:54:53Z eudoxia joined #sbcl
2015-01-22T21:00:48Z nikki93 joined #sbcl
2015-01-22T21:03:18Z nikki93 quit (Remote host closed the connection)
2015-01-22T21:04:19Z angavrilov quit (Remote host closed the connection)
2015-01-22T21:04:40Z scymtym: Krystof: sorry to bother you with this again, but can you give a rough assessment regarding the possibility of an SBCL meeting in Bielefeld? can be in irc, doesn't have be an reply-to-all email. (i was asked by my superiors and promised to ask)
2015-01-22T21:05:22Z Krystof: fair.  Won't be before Summer; unlikely as ELS will have sapped all my organizational energy
2015-01-22T21:05:32Z Krystof: sorry for non-responsiveness
2015-01-22T21:07:24Z scymtym: thanks. can i relay this as "unlikely to happen. if it does, details later."?
2015-01-22T21:07:43Z Krystof: yes
2015-01-22T21:08:42Z scymtym: speaking of ELS: would like to finish and submit the pattern-specializer work?
2015-01-22T21:09:03Z scymtym: (not on your own, i phrased that badly)
2015-01-22T21:10:26Z Krystof: unfortunately I have no capacity for extra-curricular work in the next two months at all
2015-01-22T21:10:52Z Krystof: I have been given an enormous teaching and administrative load this term, and have to run just to keep my head afloat (to mix a hilarious metaphor)
2015-01-22T21:11:16Z scymtym: sorry to hear that
2015-01-22T21:20:43Z Adlai quit (Remote host closed the connection)
2015-01-22T21:22:08Z Adlai joined #sbcl
2015-01-22T21:23:17Z nikki93 joined #sbcl
2015-01-22T21:25:16Z eudoxia quit (Quit: Leaving)
2015-01-22T21:41:36Z Shinmera quit (Quit: しつれいしなければならないんです。)
2015-01-22T21:42:18Z Shinmera joined #sbcl
2015-01-22T21:46:15Z nikki93 quit (Remote host closed the connection)
2015-01-22T22:09:17Z Adlai quit (Ping timeout: 250 seconds)
2015-01-22T22:09:39Z Adlai joined #sbcl
2015-01-22T22:12:40Z dekhoda quit (Quit: dekhoda)
2015-01-22T22:16:32Z stassats joined #sbcl
2015-01-22T22:17:37Z stassats: noticed that there's always 64 bytes used in the thread allocation region after gcing
2015-01-22T22:18:01Z stassats: turns out, 48 of those are from SB-KERNEL:RUN-PENDING-FINALIZERS allocating a closure
2015-01-22T22:18:32Z stassats: easily solvable, but then i noticed that it can also avoid consing a list of functions to run by reusing the deleted conses from *finalizer-store*
2015-01-22T22:18:34Z jasom quit (Read error: Connection reset by peer)
2015-01-22T22:18:57Z stassats: now to figure out where the 16 bytes of (nil . nil) are coming from
2015-01-22T22:20:29Z nyef: Sixteen bytes? So, one CONS?
2015-01-22T22:20:37Z stassats: yes, (nil . nil)
2015-01-22T22:21:03Z nyef: GC generation counter.
2015-01-22T22:22:21Z stassats: but isn't it allocated before the gc? or is it just not moved?
2015-01-22T22:22:40Z nyef: No idea. It's just the first thing that came to mind
2015-01-22T22:22:53Z stassats: let's put something funny instead of nil
2015-01-22T22:23:44Z stassats: that's it
2015-01-22T22:30:21Z akkad: stassats: you use sbcl on openbsd with threads?
2015-01-22T22:30:22Z akkad:  
2015-01-22T22:30:31Z stassats: i don't know?
2015-01-22T22:36:09Z Adlai quit (Ping timeout: 250 seconds)
2015-01-22T22:36:39Z Adlai joined #sbcl
2015-01-22T22:38:17Z stassats: not sure how to get rid of this cons
2015-01-22T22:38:27Z stassats: but 16 bytes instead of 64 is already better
2015-01-22T22:40:27Z gingerale quit (Ping timeout: 252 seconds)
2015-01-22T22:43:15Z nikki93 joined #sbcl
2015-01-22T22:44:11Z nikki93 quit (Remote host closed the connection)
2015-01-22T22:47:17Z stassats:  just after gc it's basically equals to the start of the thread allocation region
2015-01-22T22:47:27Z DeadTrickster quit (Ping timeout: 264 seconds)
2015-01-22T22:47:35Z nyef: Can you allocate it pre-gc and only assign it post-gc?
2015-01-22T22:47:46Z DeadTrickster joined #sbcl
2015-01-22T22:48:01Z stassats: pre-gc? well, there may be no space left
2015-01-22T22:48:07Z pkhuong_: 64 bit platforms would just use a fixnum
2015-01-22T22:48:12Z pkhuong_: 2^60 GCs?
2015-01-22T22:48:34Z pkhuong_: *could
2015-01-22T22:49:14Z stassats: why do we want to know if another gc happened or not?
2015-01-22T22:49:29Z pkhuong_: it's useful as a hint that objects may have been moved around
2015-01-22T22:49:45Z pkhuong_: useful for address-based hash tables.
2015-01-22T22:50:31Z stassats: ah, ok, i didn't see its usage outside of sub-gc
2015-01-22T22:52:43Z nyef: And, yes, lazy rehashing is its primary use AFAIK.
2015-01-22T22:53:49Z stassats: ok, 2 words are not enough to bother
2015-01-22T22:55:24Z stassats: and each new thread has 176 bytes already allocated
2015-01-22T22:56:46Z stassats: handler clusters, a cons (#<FUNCTION (LAMBDA (#:STREAM1198) :IN SB-THREAD::INITIAL-THREAD-FUNCTION-TRAMPOLINE) {100016B58B}>)
2015-01-22T22:57:28Z DeadTrickster quit (Ping timeout: 245 seconds)
2015-01-22T22:57:45Z DeadTrickster joined #sbcl
2015-01-22T22:58:51Z stassats: make that a closure
2015-01-22T23:02:18Z nyef: ... Handler clusters? Not stack-allocated?
2015-01-22T23:02:42Z nyef: That might be an easy win, actually.
2015-01-22T23:04:47Z stassats: that saves 48 bytes
2015-01-22T23:07:01Z nyef: Where's that closure coming from, anyway?
2015-01-22T23:09:47Z stassats: new conses on *handler-clusters* are DX
2015-01-22T23:09:52Z stassats: but *restart-clusters* are not DX
2015-01-22T23:11:15Z nyef: Hey, do we have any protection against some maniac marshaling a closed-over exit across thread boundaries?
2015-01-22T23:12:30Z stassats: and the closure is from with-simple-restart report function
2015-01-22T23:12:59Z stassats: wait, that shouldn't be closed over
2015-01-22T23:14:31Z stassats: so, removing restart-case and dxing clusters, the thread alloc region is not even born yet
2015-01-22T23:15:25Z nyef: Sounds good to me.
2015-01-22T23:16:00Z stassats: one more thing to look at, the closure to call the trampoline
2015-01-22T23:16:22Z stassats: it's also pinned, so, dxing == double win?
2015-01-22T23:16:45Z stassats: or not
2015-01-22T23:17:46Z stassats: the semaphores should ensure that it's still within the extent
2015-01-22T23:18:30Z Shinmera quit (Quit: しつれいしなければならないんです。)
2015-01-22T23:19:03Z attila_lendvai quit (Quit: Leaving.)
2015-01-22T23:20:07Z stassats: should probably conditionionalize pinning on :STACK-ALLOCATABLE-CLOSURES
2015-01-22T23:20:25Z stassats: although getting that to work is usually far easier than threading
2015-01-22T23:21:03Z stassats: only hppa doesn't have it
2015-01-22T23:22:31Z stassats: so, creating 1000 threads, was 673,392 bytes consed, became: 435,664 bytes consed
2015-01-22T23:23:49Z Adlai quit (Ping timeout: 250 seconds)
2015-01-22T23:26:48Z slyrus joined #sbcl
2015-01-22T23:29:41Z stassats: DXing the setup semaphore: 365,728 bytes
2015-01-22T23:30:53Z nyef: Pin should be "free" as long as you have stack-allocatable-fixed-objects.
2015-01-22T23:30:56Z stassats: waitqueue and mutex: 300,256
2015-01-22T23:32:39Z Bicyclidine quit (Ping timeout: 244 seconds)
2015-01-22T23:33:32Z Bicyclidine joined #sbcl
2015-01-22T23:34:33Z stassats: creating the thread structure takes 228,928 bytes, so, need to find out where is the reset from
2015-01-22T23:38:34Z stassats: the thread results
2015-01-22T23:40:44Z stassats: interestingly, the comment for thread-result claims that it's allocated in the parent thread, which is not true
2015-01-22T23:41:44Z stassats: could actually use two structure slots instead
2015-01-22T23:42:33Z stassats: that's enough stuff to figure out
2015-01-22T23:45:13Z stassats: actually can combine %alive-p %ephemeral-p and result-car into a single slot
2015-01-22T23:45:43Z psilord quit (Quit: Leaving.)
2015-01-22T23:56:47Z pkhuong_: DX semaphore seems like a super bad idea
2015-01-22T23:57:02Z stassats: just for the thread setup
2015-01-22T23:57:40Z stassats: the way it's used, it should stay in extent
2015-01-22T23:57:51Z pkhuong_: can the parent ever be interrupted?
2015-01-22T23:58:30Z stassats: it's under with-system-mutex (*make-thread-lock*)
2015-01-22T23:59:44Z stassats: that doesn't seem to be of the without interrupts variety