2015-05-08T00:16:47Z stassats quit (Ping timeout: 265 seconds)
2015-05-08T00:32:15Z tsumetai quit (Ping timeout: 250 seconds)
2015-05-08T00:45:10Z scymtym quit (Remote host closed the connection)
2015-05-08T00:45:28Z scymtym joined #sbcl
2015-05-08T00:57:01Z cosmicexplorer: hey so i've been playing with libclang and i've managed to bounce a c program's ast to sexp and read it into lisp to do fun things with
2015-05-08T00:57:21Z cosmicexplorer: but when i try the same thing on hello_world.cpp, reading it into lisp fails due to recursion depth constraints
2015-05-08T00:57:50Z cosmicexplorer: because (read) goes recursive when the sexp is nested
2015-05-08T00:58:46Z cosmicexplorer: i asked around on #lisp if there was a standardized version of (read) that's not recursive or at least avoids this issue
2015-05-08T00:59:14Z cosmicexplorer: i've written my own very limited version of just that but i'd like to know if there's a canonical solution to reading in massively nested sexps that recursive read chokes on
2015-05-08T01:01:08Z pkhuong: get more stack space
2015-05-08T01:02:19Z cosmicexplorer: yeah i tried using the stack space switch but it seemed to fail at the same recursion depth
2015-05-08T01:02:42Z cosmicexplorer: is there a separation of recursion depth and stack space in sbcl? maybe i was doing it wrong; i tried playing with prlimit as well in case the os wasn't letting it grow
2015-05-08T01:03:35Z pkhuong: the stack space switch is for new threads
2015-05-08T01:05:29Z cosmicexplorer: so it wouldn't affect the main thread? does that mean prlimit is the way to increase the allowed function depth?
2015-05-08T01:10:07Z pkhuong: or use the switch and spawn a new thread.
2015-05-08T01:10:20Z pkhuong: many factors are in play for the main thread's stack size
2015-05-08T01:11:53Z attila_lendvai quit (Ping timeout: 246 seconds)
2015-05-08T01:12:46Z cosmicexplorer: hm ok i wasn't aware of that
2015-05-08T01:12:51Z cosmicexplorer: i'll try that
2015-05-08T01:24:22Z brucem: pkhuong: You might find https://gist.github.com/waywardmonkeys/b0934fb1593c316e1f5e interesting.
2015-05-08T01:39:28Z scymtym quit (Remote host closed the connection)
2015-05-08T01:59:28Z rszeno quit (Quit: Leaving.)
2015-05-08T02:02:49Z Adlai quit (Ping timeout: 250 seconds)
2015-05-08T02:15:00Z echo-area joined #sbcl
2015-05-08T03:31:37Z nyef quit (Ping timeout: 264 seconds)
2015-05-08T03:35:38Z gingerale joined #sbcl
2015-05-08T03:53:48Z foom quit (Ping timeout: 256 seconds)
2015-05-08T03:53:51Z kanru quit (Ping timeout: 276 seconds)
2015-05-08T03:54:17Z dougk_ quit (Ping timeout: 265 seconds)
2015-05-08T03:56:02Z kanru joined #sbcl
2015-05-08T04:11:38Z zacts joined #sbcl
2015-05-08T04:25:00Z dougk_ joined #sbcl
2015-05-08T04:35:48Z foom joined #sbcl
2015-05-08T04:47:14Z foom quit (Read error: Connection reset by peer)
2015-05-08T05:02:07Z akkad quit (Excess Flood)
2015-05-08T05:03:28Z sdemarre joined #sbcl
2015-05-08T05:05:06Z milosn quit (Quit: leaving)
2015-05-08T05:07:13Z akkad joined #sbcl
2015-05-08T05:11:12Z oleo quit (Quit: Leaving)
2015-05-08T05:20:08Z flip214: cosmicexplorer: did you release the C=>AST=>SEXP thing somewhere, eg. on github?
2015-05-08T06:01:48Z cosmicexplorer: yeah it's on github right now it's just mega shambly since this is 1) the first time i've played with libclang and 2) i mostly do emacs lisp so i'm still janky with cl
2015-05-08T06:02:24Z cosmicexplorer: https://github.com/cosmicexplorer/semantic-code-browser i was gonna try to get it working next week
2015-05-08T06:04:07Z cosmicexplorer: if you're interested in any part of it i can put up a readme
2015-05-08T06:04:40Z sdemarre quit (Ping timeout: 272 seconds)
2015-05-08T06:06:36Z foom joined #sbcl
2015-05-08T06:10:40Z cosmicexplorer: the actual conversion to sexp actually (somehow) works just fine it's just reading it in that's currently failing
2015-05-08T06:11:44Z Quadrescence joined #sbcl
2015-05-08T06:12:39Z tsumetai joined #sbcl
2015-05-08T06:15:01Z foom quit (Read error: Connection reset by peer)
2015-05-08T06:17:40Z pranavrc joined #sbcl
2015-05-08T06:32:09Z tsumetai quit (Ping timeout: 240 seconds)
2015-05-08T06:35:54Z nikki93_ joined #sbcl
2015-05-08T06:36:58Z tsumetai joined #sbcl
2015-05-08T06:38:06Z psy_ quit (Ping timeout: 272 seconds)
2015-05-08T06:53:22Z flip214: cosmicexplorer: thanks, looking right now.
2015-05-08T06:53:49Z flip214: well, I thought that you're using Clang from *within* CL - then there wouldn't be the serialization issues.
2015-05-08T07:04:12Z rszeno joined #sbcl
2015-05-08T07:04:38Z Shinmera joined #sbcl
2015-05-08T07:09:52Z rszeno quit (Quit: Leaving.)
2015-05-08T07:16:41Z angavrilov joined #sbcl
2015-05-08T07:19:43Z foom joined #sbcl
2015-05-08T07:27:05Z ams quit (Quit: bz)
2015-05-08T07:27:49Z Cymew joined #sbcl
2015-05-08T07:39:58Z fikusz quit (Ping timeout: 244 seconds)
2015-05-08T07:42:05Z edgar-rft quit (Quit: edgar-rft)
2015-05-08T07:42:20Z edgar-rft joined #sbcl
2015-05-08T07:42:35Z edgar-rft quit (Remote host closed the connection)
2015-05-08T07:43:04Z Quadrescence quit (Quit: This computer has gone to sleep)
2015-05-08T07:43:35Z cosmicexplorer: ohhhhhh that would be actually not a bad idea
2015-05-08T07:46:18Z cosmicexplorer: i recall the sbcl ffi being really chill last time i used it that could totally work
2015-05-08T07:47:05Z flip214: if you can interface with the c++ functions, you could drop walk-ast.cpp
2015-05-08T07:50:02Z cosmicexplorer: exactly and that'd be lovely
2015-05-08T07:50:05Z jlarocco quit (Quit: This computer has gone to sleep)
2015-05-08T07:50:34Z jlarocco joined #sbcl
2015-05-08T07:50:44Z cosmicexplorer: i'm interning for microsoft in a week and i'm gonna gag on the sheer lines of c++ i'll be forced to write better not to exacerbate the issue
2015-05-08T07:51:26Z jlarocco quit (Client Quit)
2015-05-08T07:53:34Z jlarocco joined #sbcl
2015-05-08T07:57:52Z brucem: libclang has a C interface.
2015-05-08T07:58:26Z cosmicexplorer: yeah there are c++ bindings too but they complicate everything so i'm just using c
2015-05-08T08:01:05Z fikusz joined #sbcl
2015-05-08T08:03:55Z nikki93_ quit (Remote host closed the connection)
2015-05-08T08:04:14Z nikki93_ joined #sbcl
2015-05-08T08:05:52Z nikki93_ quit (Remote host closed the connection)
2015-05-08T08:13:35Z White_Flame joined #sbcl
2015-05-08T08:33:21Z tsumetai quit (Ping timeout: 276 seconds)
2015-05-08T08:44:34Z tsumetai joined #sbcl
2015-05-08T09:35:51Z Shinmera quit (Quit: しつれいしなければならないんです。)
2015-05-08T09:57:38Z gabriel_laddel joined #sbcl
2015-05-08T09:57:59Z gabriel_laddel quit (Changing host)
2015-05-08T09:58:00Z gabriel_laddel joined #sbcl
2015-05-08T10:05:24Z scymtym joined #sbcl
2015-05-08T10:20:06Z echo-area quit (Remote host closed the connection)
2015-05-08T10:26:26Z Shinmera joined #sbcl
2015-05-08T10:32:32Z tsumetai quit (Ping timeout: 244 seconds)
2015-05-08T10:35:40Z Adlai joined #sbcl
2015-05-08T10:40:13Z nyef joined #sbcl
2015-05-08T11:02:56Z salv0 quit (Remote host closed the connection)
2015-05-08T11:12:11Z attila_lendvai joined #sbcl
2015-05-08T11:12:55Z tsumetai joined #sbcl
2015-05-08T11:19:24Z White_Flame quit (Ping timeout: 245 seconds)
2015-05-08T11:44:27Z stassats joined #sbcl
2015-05-08T12:06:56Z pranavrc quit (Ping timeout: 240 seconds)
2015-05-08T12:11:33Z pranavrc joined #sbcl
2015-05-08T12:13:46Z pkhuong: clang and llvm are really bad at embedding
2015-05-08T12:14:00Z pkhuong: they assert a llot and don't expect to recover from sigabrt
2015-05-08T12:19:22Z pkhuong: brucem: untagging with xorq   $0x1, %rax seems suboptimal
2015-05-08T12:19:56Z pkhuong: of course, the right way to do this is to tag fixnums with 0 ;)
2015-05-08T12:20:28Z pkhuong: how does xorq   $0x1, %rax know what value overflowed?
2015-05-08T12:20:40Z brucem: pkhuong: It just does that to clear the tag on the 2nd value, so it can add them in tagged state (as you know) ... we use tag 0 for pointers for better or worse.
2015-05-08T12:21:10Z brucem: pkhuong: the overflow happens (or not) in the next line, the addq
2015-05-08T12:21:55Z pkhuong: sorry, bad copy paste how does Kmachine_word_overflowVKmI know what value overflowed
2015-05-08T12:24:04Z pkhuong: if you untag with a subtraction, you can use inc/dec on x86 and LLVM will be better able to simplify arithmetic
2015-05-08T12:24:37Z brucem: pkhuong: since we aren't upgrading to a bignum, we don't have to know the exact value ... for the most part, we just need the right operation to fail (which it does here)
2015-05-08T12:26:16Z brucem: pkhuong: we also have an int overflow signal handler (https://github.com/dylan-lang/opendylan/blob/master/sources/lib/run-time/llvm-exceptions.c#L135) ... but now i'm not sure where that would get triggered.
2015-05-08T12:26:27Z pkhuong: the codegen in the epilogue for returning values is really weird
2015-05-08T12:27:04Z pkhuong: is it hand rolled in the calling convention module?
2015-05-08T12:27:35Z pkhuong: but yeah, pretty cool integration
2015-05-08T12:27:43Z brucem: pkhuong: he uses fastcc and teh struct MV return value.
2015-05-08T12:28:01Z brucem: pkhuong: we see a huge cost from using TLS to store an MV count and this is (significantly) faster.
2015-05-08T12:28:42Z nyef: Let me guess, all your TLS is pretty much full-calls?
2015-05-08T12:29:21Z brucem: nyw: shared libraries are involved, so ...
2015-05-08T12:29:26Z brucem: err, nyef: ^
2015-05-08T12:29:48Z brucem: pkhuong: It is early days for this ... although he's been at it for 5+ years off and on.
2015-05-08T12:32:13Z brucem: pkhuong: and it is already significantly better than our C back-end ... and source level debugging via DWARF is mostly working.
2015-05-08T12:32:56Z tsumetai quit (Ping timeout: 240 seconds)
2015-05-08T12:33:08Z pkhuong: brucem: I think I kind of see what's going on.
2015-05-08T12:33:30Z pkhuong: the xor is messing up llvm's detection of commutativity
2015-05-08T12:34:16Z pkhuong: it does r14 <- rax + r14; r14 -> rbx; rbx -> rax
2015-05-08T12:34:49Z pkhuong: ideally, you'd get rax <- rax + r14
2015-05-08T12:36:11Z pkhuong: the mov r14, rbx / movb 1, dl / mov rbx, rax sequence at 0x100003e78 is still unexpected.
2015-05-08T12:37:00Z tsumetai joined #sbcl
2015-05-08T12:37:30Z brucem: pkhuong: if you look at the LLVM IR, I think itll be more clear.
2015-05-08T12:37:58Z brucem: pkhuong: We return struct { value, mv_count } MV ... and then any MV after the first are available elsewhere. But the common case is fast.
2015-05-08T12:38:29Z pkhuong: I understand what you're returning
2015-05-08T12:38:42Z brucem: pkhuong: the odd thing is that I don't know where we untag via xor to test out changing it :)
2015-05-08T12:38:54Z pkhuong: what I'm sayig is that the regalloc is weak, and that the codegen exacerbates the rest
2015-05-08T12:39:07Z karswell joined #sbcl
2015-05-08T12:52:22Z brucem: pkhuong: I'm currently mystified as to where that xor is coming from. :/
2015-05-08T12:53:36Z brucem: I just had to say that and I found it.
2015-05-08T12:55:32Z oleo joined #sbcl
2015-05-08T12:58:12Z psilord quit (Quit: Leaving.)
2015-05-08T13:05:41Z brucem: pkhuong: never as easy as it might seem ... this is interesting. I'm going to have a file a bug to track it.
2015-05-08T13:05:44Z brucem: thanks!
2015-05-08T13:05:48Z brucem: I love rabbit holes.
2015-05-08T13:06:02Z brucem: I spend 80% of my time in them.
2015-05-08T13:06:11Z oleo: lol
2015-05-08T13:06:32Z brucem: oleo: no deadline pressure => do what seems interesting => often leads to rabbit holes.
2015-05-08T13:06:54Z oleo: jep
2015-05-08T13:07:09Z oleo: for working anyway, for learning you can't circumvent that i think.....
2015-05-08T13:08:54Z brucem: oleo: my work is compiler and run-time support ... lots of rabbit holes there too.
2015-05-08T13:09:20Z oleo: hmmmm
2015-05-08T13:10:46Z oleo: i have to port some of fatemans code to sbcl i think.....
2015-05-08T13:10:47Z oleo: meh
2015-05-08T13:12:24Z oleo: it's gcl only most of the time....
2015-05-08T13:14:20Z pranavrc quit
2015-05-08T13:14:39Z edgar-rft joined #sbcl
2015-05-08T13:23:26Z karswell quit (Ping timeout: 272 seconds)
2015-05-08T13:38:09Z foom quit (Ping timeout: 265 seconds)
2015-05-08T13:39:32Z selat joined #sbcl
2015-05-08T13:51:07Z foom joined #sbcl
2015-05-08T13:53:48Z brucem: pkhuong: https://github.com/dylan-lang/opendylan/issues/897 ... changing from xor to sub in the LLVM IR and recompiling that function doesn't result in better register usage. :(
2015-05-08T13:56:16Z psilord joined #sbcl
2015-05-08T13:58:50Z nyef: How about... using 0 for the fixnum tag?
2015-05-08T14:04:32Z salv0 joined #sbcl
2015-05-08T14:05:47Z brucem: nyef: Maybe one day. But not this year unless someone new were to step up.
2015-05-08T14:07:18Z stassats: bloody cold-init errors
2015-05-08T14:10:37Z Cymew quit (Ping timeout: 264 seconds)
2015-05-08T14:22:35Z pkhuong: brucem: it does however save 3-4 bytes of instruction encoding
2015-05-08T14:32:56Z tsumetai quit (Ping timeout: 240 seconds)
2015-05-08T14:34:49Z tsumetai` joined #sbcl
2015-05-08T14:36:17Z brucem: pkhuong: I'm testing a fix for it with a 3 stage bootstrap.
2015-05-08T14:37:19Z salv0 quit (Remote host closed the connection)
2015-05-08T14:48:54Z stassats: huh, no %map transform results in a memfault
2015-05-08T15:10:46Z zacts quit (Quit: ERC (IRC client for Emacs 24.5.1))
2015-05-08T15:13:03Z stassats: huh, replacing (map 'simple-string #'code-char item) with (map-into (make-string (length item)) #'code-char item) results in a not-a-simple-vector error
2015-05-08T15:30:18Z scymtym quit (Read error: Connection reset by peer)
2015-05-08T16:03:35Z dougk_ quit (Remote host closed the connection)
2015-05-08T16:03:55Z dougk_ joined #sbcl
2015-05-08T16:16:34Z scymtym joined #sbcl
2015-05-08T16:28:47Z hlavaty quit (Remote host closed the connection)
2015-05-08T16:32:11Z zacts joined #sbcl
2015-05-08T16:33:09Z tsumetai` quit (Ping timeout: 245 seconds)
2015-05-08T16:42:30Z stassats: why OPEN's element-type defaults to base-char?
2015-05-08T16:48:43Z stassats: why does (sb-sys:%primitive print "abc")
2015-05-08T16:48:47Z stassats: only print "a"?
2015-05-08T16:48:58Z sdemarre joined #sbcl
2015-05-08T16:49:19Z nyef: %primitive print is only defined to work on base-strings.
2015-05-08T16:49:35Z stassats: just realized that
2015-05-08T16:49:56Z stassats: why doesn't print check the type?
2015-05-08T16:50:35Z nyef: It's meant as a low-level debugging aid... and pre-dates (ARRAY NIL).
2015-05-08T16:51:32Z stassats: so, memory faults are better than checking types?
2015-05-08T16:52:05Z nyef: So almost nobody uses it outside of the cross-compiler, which ONLY produces base-strings.
2015-05-08T16:52:56Z nyef: Basically, what I'm saying is that it's this way because of the historical context, not for any especially critical reason.
2015-05-08T16:55:40Z zacts quit (Ping timeout: 246 seconds)
2015-05-08T16:58:25Z stassats: so, is (map 'simple-string #'code-char item) equivalent to (map-into (make-string (length item)) #'code-char item) or am i blind?
2015-05-08T17:01:37Z Xof: there is some magic regarding string type specifiers, but I can't remember whether it applies here.  (map 'simple-string ...) needs that magic, because otherwise it doesn't specify the element type fully
2015-05-08T17:02:14Z stassats: but somehow ends up complaining about a string not being simple-vector
2015-05-08T17:02:48Z dougk_: would (map 'simple-string ..) be permitted to determine that all characters were base-char, and actually make a base-string?
2015-05-08T17:03:00Z Xof: no
2015-05-08T17:03:03Z Xof: I've just found the magic
2015-05-08T17:03:07Z Xof: http://www.lispworks.com/documentation/lw51/CLHS/Body/t_smp_st.htm
2015-05-08T17:03:29Z Xof: so, stassats, I think they should be equivalent
2015-05-08T17:03:36Z dougk_: also two comments: (1) you never need to use %primitive print any more except on Windows. PRINT works immediately in cold-init.
2015-05-08T17:04:19Z stassats: what's the problem with windows?
2015-05-08T17:04:36Z dougk_: I didn't figure out how to hook into a windows stream right away
2015-05-08T17:05:00Z dougk_: on Unix i have a character-at-a-time output thing to fd 2. It's that simple, and, yes, incredibly inefficient
2015-05-08T17:05:27Z dougk_: write-string goes as one chunk, but writing each parenthesis in a list is a system call.
2015-05-08T17:05:34Z dougk_: until the real buffered streams are set up
2015-05-08T17:06:27Z dougk_: .. and (2) I think my patch stood to be slightly more efficient than map-into with the addition of the 'fixme' that I added, which is that we *know* that I calculated the correct output length, so the termination condition is precise
2015-05-08T17:06:39Z dougk_: map-into still has to go the shorter of the output, or the shortest input.
2015-05-08T17:07:07Z dougk_: and my patch fundamentally relies on the *same* code so it's not as if I reinvented too much.
2015-05-08T17:07:25Z stassats: map-into and make-sequence is simpler
2015-05-08T17:07:42Z stassats: except that it gets miscompiled by xc
2015-05-08T17:09:33Z gingerale quit (Ping timeout: 256 seconds)
2015-05-08T17:10:11Z Xof: let's see your transform?
2015-05-08T17:10:41Z stassats: that's no transform at all
2015-05-08T17:15:22Z dougk_:  'setf.lisp' is almost devoid of known bugs after 15 commits :-)
2015-05-08T17:18:19Z karswell joined #sbcl
2015-05-08T17:26:08Z stassats: vector-map-into has safety 0, maybe it confuses the error
2015-05-08T17:30:10Z stassats: replacing it with debug doesn't help
2015-05-08T17:39:16Z stassats` joined #sbcl
2015-05-08T17:39:35Z stassats`: trace-file doesn't show anything interesting
2015-05-08T17:40:17Z stassats quit (Ping timeout: 246 seconds)
2015-05-08T17:45:03Z stassats`: so, the problem is with full-calling map-into
2015-05-08T17:49:32Z stassats`: so, sb!unicode::!unicode-properties-cold-init is called before the top-level cold-init forms are called
2015-05-08T17:49:45Z stassats`: and %%VECTOR-MAP-INTO-FUNS%% is initialized only then
2015-05-08T17:49:47Z cosmicexplorer quit (Ping timeout: 256 seconds)
2015-05-08T17:53:13Z stassats`: what a stupid waste of time
2015-05-08T17:57:41Z mega1 joined #sbcl
2015-05-08T18:00:09Z davazp joined #sbcl
2015-05-08T18:06:44Z dougk_: That vector can be initialized during cold-load, not cold-init.  Should fix your problem.
2015-05-08T18:11:33Z stassats`: and transforming into map-into is better if the transform doesn't go through, since it can better process the result type
2015-05-08T18:11:38Z stassats`: without parsing it at run-time
2015-05-08T18:14:54Z cosmicexplorer joined #sbcl
2015-05-08T18:31:33Z oleo_ joined #sbcl
2015-05-08T18:33:21Z stassats`: dougk_: how exactly can i do that?
2015-05-08T18:33:36Z dougk_: stassats: I'm working on that for you.
2015-05-08T18:34:07Z dougk_: basically we're going to need a few things: the (defglobal) emitted by define-array-dispatch needs to become !defglobal.
2015-05-08T18:34:20Z dougk_: But !defglobal can't execute make-array (yet), so i have to teach it that.
2015-05-08T18:34:36Z oleo quit (Ping timeout: 252 seconds)
2015-05-08T18:34:53Z dougk_: then we have to get the (setf) for each element to become cold-init-forms, probably.
2015-05-08T18:39:05Z gingerale joined #sbcl
2015-05-08T18:39:06Z nyef: ... warmer cold-cores?
2015-05-08T18:39:24Z dougk_: I've been pre-warming the cold-core since forever ago
2015-05-08T18:39:31Z nyef: So it seems.
2015-05-08T18:39:52Z dougk_: makes things so much easier to reason about during cold-init
2015-05-08T18:40:01Z dougk_: (when it works)
2015-05-08T18:43:21Z gingerale quit (Remote host closed the connection)
2015-05-08T18:57:24Z stassats`: map-into should probably also transform to these specialized functions, when the input sequences are not known
2015-05-08T18:59:13Z stassats`: not sure how to convince map-into not to do the length checks
2015-05-08T19:08:56Z tsumetai` joined #sbcl
2015-05-08T19:20:13Z attila_lendvai quit (Ping timeout: 264 seconds)
2015-05-08T19:27:40Z cosmicexplorer quit (Ping timeout: 256 seconds)
2015-05-08T19:28:51Z cosmicexplorer joined #sbcl
2015-05-08T19:33:29Z dougk_ quit (Ping timeout: 256 seconds)
2015-05-08T19:38:02Z prxq joined #sbcl
2015-05-08T19:46:55Z dougk_ joined #sbcl
2015-05-08T19:50:15Z nikki93_ joined #sbcl
2015-05-08T20:07:01Z davazp quit (Ping timeout: 255 seconds)
2015-05-08T20:18:37Z ASau quit (Remote host closed the connection)
2015-05-08T20:19:25Z ASau joined #sbcl
2015-05-08T20:21:54Z davazp joined #sbcl
2015-05-08T20:30:59Z selat quit (Quit: Lost terminal)
2015-05-08T20:33:24Z tsumetai` quit (Ping timeout: 256 seconds)
2015-05-08T20:34:52Z sdemarre quit (Ping timeout: 252 seconds)
2015-05-08T21:04:18Z angavrilov quit (Remote host closed the connection)
2015-05-08T21:27:54Z prxq quit (Remote host closed the connection)
2015-05-08T21:32:49Z edgar-rft quit (Quit: edgar-rft)
2015-05-08T21:39:25Z tsumetai joined #sbcl
2015-05-08T22:03:03Z zacts joined #sbcl
2015-05-08T22:03:24Z zacts quit
2015-05-08T22:06:06Z nikki93_ quit (Remote host closed the connection)
2015-05-08T22:20:02Z cosmicexplorer quit (Ping timeout: 264 seconds)
2015-05-08T22:32:43Z psilord quit (Quit: Leaving.)
2015-05-08T22:32:48Z tsumetai` joined #sbcl
2015-05-08T22:33:41Z tsumetai quit (Ping timeout: 265 seconds)
2015-05-08T22:44:34Z davazp quit (Read error: Connection reset by peer)
2015-05-08T23:21:44Z nikki93_ joined #sbcl
2015-05-08T23:23:55Z nikki93_ quit (Remote host closed the connection)
2015-05-08T23:43:03Z Shinmera quit (Quit: しつれいしなければならないんです。)
2015-05-08T23:56:27Z psilord joined #sbcl