2014-12-25T00:20:55Z nikki93 quit (Remote host closed the connection)
2014-12-25T01:23:46Z michael_lee joined #sbcl
2014-12-25T01:42:36Z michael_lee quit (Ping timeout: 256 seconds)
2014-12-25T01:46:45Z ehaliewicz quit (Ping timeout: 272 seconds)
2014-12-25T02:17:09Z echo-area joined #sbcl
2014-12-25T02:59:04Z edgar-rft quit (Quit: happening destroyed into perpetual bleeding)
2014-12-25T03:39:27Z christoph_debian quit (Ping timeout: 245 seconds)
2014-12-25T03:52:32Z christoph_debian joined #sbcl
2014-12-25T04:16:57Z ASau quit (Ping timeout: 245 seconds)
2014-12-25T04:26:29Z scymtym quit (Ping timeout: 245 seconds)
2014-12-25T04:56:17Z nikki93 joined #sbcl
2014-12-25T05:01:01Z nikki93 quit (Ping timeout: 252 seconds)
2014-12-25T05:20:58Z Posterdati quit (Ping timeout: 255 seconds)
2014-12-25T05:21:26Z Posterdati joined #sbcl
2014-12-25T05:58:26Z nyef quit (Quit: G'night all)
2014-12-25T06:41:33Z milosn joined #sbcl
2014-12-25T06:42:13Z psy_ quit (Quit: Leaving)
2014-12-25T07:13:48Z psy joined #sbcl
2014-12-25T07:33:51Z nikki93 joined #sbcl
2014-12-25T08:00:44Z nikki93 quit (Remote host closed the connection)
2014-12-25T08:01:06Z nikki93 joined #sbcl
2014-12-25T08:12:36Z stassats joined #sbcl
2014-12-25T08:31:37Z _8hzp quit (Ping timeout: 240 seconds)
2014-12-25T08:43:04Z hzp joined #sbcl
2014-12-25T08:52:38Z oleo joined #sbcl
2014-12-25T08:54:37Z oleo__ quit (Ping timeout: 240 seconds)
2014-12-25T09:09:38Z pppp2 joined #sbcl
2014-12-25T09:14:52Z milosn quit (Ping timeout: 245 seconds)
2014-12-25T09:51:04Z milosn joined #sbcl
2014-12-25T10:02:53Z Bicyclidine quit (Ping timeout: 244 seconds)
2014-12-25T10:29:38Z Bicyclidine joined #sbcl
2014-12-25T10:42:46Z angavrilov joined #sbcl
2014-12-25T10:53:00Z echo-area quit (Remote host closed the connection)
2014-12-25T11:01:06Z stassats` joined #sbcl
2014-12-25T11:05:18Z stassats quit (Ping timeout: 256 seconds)
2014-12-25T11:23:37Z DeadTrickster quit (Ping timeout: 245 seconds)
2014-12-25T11:51:05Z DeadTrickster joined #sbcl
2014-12-25T12:34:29Z echo-area joined #sbcl
2014-12-25T12:56:57Z scymtym joined #sbcl
2014-12-25T13:12:24Z milosn quit (Remote host closed the connection)
2014-12-25T13:28:28Z eudoxia joined #sbcl
2014-12-25T14:13:36Z edgar-rft joined #sbcl
2014-12-25T14:33:00Z Bicyclidine quit (Ping timeout: 258 seconds)
2014-12-25T14:46:30Z nyef joined #sbcl
2014-12-25T14:53:37Z pppp2 quit (Read error: Connection reset by peer)
2014-12-25T14:56:10Z stassats`: maybe i should try using qemu mips emulation again
2014-12-25T14:56:41Z nyef: As an emulator author, my rule is "never trust an emulator".
2014-12-25T14:57:11Z stassats`: well, it started crashing left and right for me, so i abandoned it
2014-12-25T14:58:08Z stassats`: using portable code for argchecking might be nice, but it's just much easier to extend verify-arg-count to do max and min checking
2014-12-25T14:58:57Z stassats`: at least when you can compile it
2014-12-25T14:59:43Z Bicyclidine joined #sbcl
2014-12-25T15:01:00Z stassats`: because make-xep-lambda-expression happens at ir1 time and i need to carry it until ir2
2014-12-25T15:12:54Z stassats`: if i can just get to run the sbcl.org binaries, it would be enough to test the vops
2014-12-25T15:13:47Z nyef: Cross-build?
2014-12-25T15:14:32Z nyef: I don't know how anyone managed to build a 1.0.23 or 1.0.28 MIPS build, anyway.
2014-12-25T15:16:13Z stassats`: that new mips board sounds promising
2014-12-25T15:16:20Z nyef: I couldn't build a MIPS version prior to partway through the 1.1.2 cycle, when I fixed call_into_lisp().
2014-12-25T15:16:22Z nyef: Which one?
2014-12-25T15:16:33Z stassats`: the new one
2014-12-25T15:18:02Z nyef: The "Creator CI20", or some other one?
2014-12-25T15:18:10Z stassats`: sounds like that
2014-12-25T15:19:19Z nyef: Looks to be 32-bit, but otherwise decent.
2014-12-25T15:20:09Z eudoxia quit (Quit: Leaving)
2014-12-25T15:20:55Z nyef: Dual-core, which means that it can be a bit weak with respect to testing for multiprocessing stability.
2014-12-25T15:21:48Z nyef: (The two cores could share a cache, which is not nearly as strong a guarantee as two entirely separate CPUs in the same box, let alone having two SGI nodes with a numalink between them.)
2014-12-25T15:23:15Z stassats`: and it doesn't want to boot
2014-12-25T15:25:07Z stassats`: ok, it finally did start up
2014-12-25T15:25:23Z stassats`: Linux debian-mips 3.2.0-4-5kc-malta #1 Debian 3.2.51-1 mips64 GNU/Linux
2014-12-25T15:25:28Z stassats`: ls
2014-12-25T15:26:12Z stassats`: wrong focus
2014-12-25T15:27:16Z stassats`: console=tty0 was bogus
2014-12-25T15:28:15Z nyef: At the rate I'm going, I'll have a CI20 in hand before I get my IP35 to boot anything other than IRIX. /-:
2014-12-25T15:28:51Z stassats`: so slow, so slow
2014-12-25T15:29:17Z nyef: That's the other reason I don't use it.
2014-12-25T15:29:22Z nyef: (qemu, that is)
2014-12-25T15:31:54Z stassats`: well, if it works, i can wait
2014-12-25T15:34:01Z BitPuffin joined #sbcl
2014-12-25T15:35:21Z BitPuffin: which subsystems of sbcl are BSD licensed?
2014-12-25T15:47:34Z milosn joined #sbcl
2014-12-25T15:53:10Z pkhuong: BitPuffin: all of them, except the parts that are public domain.
2014-12-25T15:54:55Z BitPuffin: pkhuong: I thought the majoryity were public domain
2014-12-25T15:55:00Z BitPuffin: majority*
2014-12-25T16:04:23Z BitPuffin: so cmucl is more in the pd?
2014-12-25T16:08:25Z nikki93 quit (Remote host closed the connection)
2014-12-25T16:09:05Z nikki93 joined #sbcl
2014-12-25T16:11:04Z pkhuong: no you're right. it's the other way around
2014-12-25T16:11:15Z pkhuong: there's a couple authors in jurisdictions that don't allow PD
2014-12-25T16:11:38Z pkhuong: if you care, it's probably best to assume BSD, because I don't know if we ever kept track of all that
2014-12-25T16:11:48Z BitPuffin: hm
2014-12-25T16:11:56Z BitPuffin: why wouldn't they use something like CC0 instead?
2014-12-25T16:12:03Z BitPuffin: since that works in those places as well
2014-12-25T16:12:09Z BitPuffin: such as I dunno, japan
2014-12-25T16:13:13Z nyef: CC0 wasn't available at the time.
2014-12-25T16:13:20Z BitPuffin: I see
2014-12-25T16:13:24Z BitPuffin: that makes sense
2014-12-25T16:13:38Z nyef: SBCL was first released in '99, IIRC.
2014-12-25T16:13:45Z nikki93 quit (Ping timeout: 264 seconds)
2014-12-25T16:13:54Z BitPuffin: I don't really care because I intend to do anything that BSD would stop me from doing I mainly care because seeing things that aren't copyrighted makes me happy :)
2014-12-25T16:14:41Z nyef: And CMUCL is older, and its parent system even older.
2014-12-25T16:15:08Z BitPuffin: cmucl was based on openmcl or something right?
2014-12-25T16:15:18Z stassats`: no
2014-12-25T16:15:26Z stassats`: that's like 20 years apart
2014-12-25T16:16:16Z BitPuffin: or was openmcl related to ccl perhaps
2014-12-25T16:16:18Z BitPuffin: I don't remember :P
2014-12-25T16:16:42Z nyef: clozure common lisp is a rebadged openmcl.
2014-12-25T16:17:19Z stassats`: which is a rebadged mcl, which is a rebadged coral common lisp
2014-12-25T16:17:39Z BitPuffin: quite the history in other words
2014-12-25T16:18:28Z BitPuffin: has there ever been any discussions to move from bsd to cc0?
2014-12-25T16:21:50Z BitPuffin: though I'm guessing that's quite a bit of work since you have to get the blessing of everyone who made bsd licensed contributions and they may not be possible to get in touch wit hetc
2014-12-25T16:24:12Z nyef: As a simple example there, if ths made any BSD-licensed contributions, then he's a bit beyond reach.
2014-12-25T16:24:25Z eudoxia joined #sbcl
2014-12-25T16:25:20Z stassats`: that went dark
2014-12-25T16:25:30Z pkhuong: did not expect
2014-12-25T16:26:32Z nyef: Yeah, sorry. Probably not quite appropriate for the day.
2014-12-25T16:26:42Z nyef: Memento Mori, huh?
2014-12-25T16:27:18Z stassats` isn't quite sure why aren't the arguments being connected in SB-C::INIT-XEP-ENVIRONMENT in case of (eq (functional-kind fun) :toplevel)
2014-12-25T16:29:28Z BitPuffin: nyef: ths?
2014-12-25T16:29:33Z BitPuffin: someone who died? :/
2014-12-25T16:30:32Z stassats`: got 1.0.23-mips running again, but i don't want to attempt a build, even with crossing
2014-12-25T16:31:06Z nyef: stassats`: I still don't understand how 1.0.23-mips got built, given the damage in mips-assem.S at the time.
2014-12-25T16:31:32Z nyef: (Consider 7d986ea35d26bbde3d8df8cc2580905b918aefef )
2014-12-25T16:32:02Z stassats`: who built it?
2014-12-25T16:32:21Z nyef: I have no idea.
2014-12-25T16:33:01Z stassats`: ths...
2014-12-25T16:33:16Z stassats`: i kind expected that
2014-12-25T16:34:02Z nyef: Yeah, ths was one of the big driving forces for open source MIPS stuff.
2014-12-25T16:34:11Z BitPuffin: nyef: either way, with the people who are reachable who makes such contributions it would be easier, just thinking that cc0 is more in the spirit of the majority of the system so it could (or could not) be worth doing I dunno :)
2014-12-25T16:37:17Z stassats`: i need to figure the whole journey of XEPs
2014-12-25T16:37:59Z stassats`: one code object can have multiple xeps, right?
2014-12-25T16:38:23Z nyef: Oh yes.
2014-12-25T16:38:36Z nyef: Here's a simple way to cause it: Return a closure.
2014-12-25T16:41:21Z stassats`: still not clear what's the special deal with :toplevel
2014-12-25T16:43:00Z stassats`: maybe if i manage to get a hold of it
2014-12-25T16:43:33Z pkhuong: isn't toplevel the kind for XEPs that are only called once, without any argument?
2014-12-25T16:43:56Z stassats`: (let () (declare)) is one
2014-12-25T16:44:48Z pkhuong: right, so it's a toplevel expression that is called for effects
2014-12-25T16:44:58Z stassats`: and which isn't fopcompiled
2014-12-25T16:46:32Z stassats`: ok, i see why it's treated specially, it has empty lambda-vars
2014-12-25T16:46:44Z stassats`: ordinary LAMBDA will have arg-count as the first arg
2014-12-25T16:47:11Z stassats`: so, the proper check would probably be (when (lambda-vars fun) instead
2014-12-25T16:47:51Z pkhuong: or assert that the two predicates are equivalent?
2014-12-25T16:49:10Z stassats`: i don't need to pass args for all required arguments anymore
2014-12-25T16:52:10Z stassats`: maybe i can rework the whole thing so that optional dispatch doesn't need to move any arguments around
2014-12-25T16:52:24Z stassats`: but that might be difficult with a lot of keys, since it may need to use some stack space
2014-12-25T16:55:03Z stassats`: and init-forms can have arbitrary code
2014-12-25T17:08:02Z stassats`: what if there were a restart for too much arguments: ignore the extra ones
2014-12-25T17:08:16Z stassats`: or adding missing arguments
2014-12-25T17:08:28Z pkhuong: that would be really useful
2014-12-25T17:12:16Z stassats`: too bad you can't reconstruct the real arguments further down the line
2014-12-25T17:13:19Z stassats`: setf, &key defaulting ruins them
2014-12-25T17:17:24Z stassats`: real arguments for arg-count-error would be a good start
2014-12-25T17:31:17Z nyef quit (Quit: Time for me to go)
2014-12-25T18:25:03Z eudoxia quit (Quit: Leaving)
2014-12-25T18:39:18Z drichards: bah, it's been to long.  back to AMOP
2014-12-25T18:51:40Z Krystof quit (Ping timeout: 256 seconds)
2014-12-25T18:58:13Z stassats` quit (Remote host closed the connection)
2014-12-25T19:07:58Z stassats joined #sbcl
2014-12-25T19:16:47Z nikki93 joined #sbcl
2014-12-25T19:19:35Z nikki93_ joined #sbcl
2014-12-25T19:23:48Z nikki93 quit (Ping timeout: 272 seconds)
2014-12-25T19:47:59Z krzysz00 quit (Quit: leaving)
2014-12-25T19:51:45Z nikki93_ quit (Remote host closed the connection)
2014-12-25T19:52:21Z nikki93 joined #sbcl
2014-12-25T19:55:45Z milosn quit (Ping timeout: 264 seconds)
2014-12-25T19:56:28Z nikki93_ joined #sbcl
2014-12-25T19:56:51Z nikki93 quit (Ping timeout: 244 seconds)
2014-12-25T19:57:32Z nikki93_ quit (Remote host closed the connection)
2014-12-25T19:58:06Z nikki93 joined #sbcl
2014-12-25T19:59:50Z nikki93 quit (Read error: No route to host)
2014-12-25T20:00:11Z nikki93 joined #sbcl
2014-12-25T20:21:51Z daviid joined #sbcl
2014-12-25T20:24:48Z milosn joined #sbcl
2014-12-25T20:55:20Z oleo quit (Quit: Verlassend)
2014-12-25T21:01:51Z oleo joined #sbcl
2014-12-25T21:08:33Z krzysz00 joined #sbcl
2014-12-25T21:15:22Z oleo quit (Remote host closed the connection)
2014-12-25T21:26:53Z oleo joined #sbcl
2014-12-25T21:42:35Z akkad: disassemble on sbcl seems much more optimized than even the commerical offerings
2014-12-25T21:42:43Z akkad: CPI is much lower
2014-12-25T21:43:21Z stassats: CPI?
2014-12-25T21:44:38Z akkad: cycles per instruction
2014-12-25T21:45:05Z stassats: that's a CPU metric
2014-12-25T21:45:34Z akkad: uh huh
2014-12-25T21:46:07Z stassats: and is constant for all implementations on the same CPU
2014-12-25T21:46:20Z oleo is now known as Guest5801
2014-12-25T21:46:37Z stassats: so, you're thinking some other term
2014-12-25T21:48:36Z oleo__ joined #sbcl
2014-12-25T21:48:40Z oleo__ quit (Read error: Connection reset by peer)
2014-12-25T21:48:57Z Guest5801 quit (Ping timeout: 240 seconds)
2014-12-25T21:48:57Z akkad: well IC * CPI is lower
2014-12-25T21:49:08Z akkad: so yes technically it's lower IC
2014-12-25T21:50:29Z stassats: if SBCL is better than commercial offerings, then those commercials offering must be pretty bad
2014-12-25T21:50:35Z oleo joined #sbcl
2014-12-25T21:50:43Z stassats: SBCL output is far from optimal
2014-12-25T21:51:10Z stassats: no proper instruction selection, not so optimal register allocation, no peephole optimization
2014-12-25T22:01:19Z nikki93 quit (Remote host closed the connection)
2014-12-25T22:25:20Z DeadTrickster: after run-program can new process access fds opened by parent?
2014-12-25T22:25:30Z DeadTrickster: I'm getting bad file descriptor
2014-12-25T22:26:27Z DeadTrickster: maybe it sets fd_cloexec somewhere?
2014-12-25T22:30:16Z hzp: could be that we're closing all non-standard file descriptors in spawn explicitly
2014-12-25T22:31:17Z DeadTrickster: fds created by pipe call if that matters
2014-12-25T22:32:17Z DeadTrickster: if I just use fork everything works, after moving to run program bad file descriptors appears
2014-12-25T22:33:51Z angavrilov quit (Remote host closed the connection)
2014-12-25T22:35:44Z DeadTrickster: *close-in-parent* looks suspicious
2014-12-25T22:43:43Z akkad: sbcl vs lispworks in disassemble of many of the ironclad functions seem to be between 1/2 and 2/3 of the lispworks version
2014-12-25T22:43:56Z stassats: ironclad is specifically optimized for sbcl
2014-12-25T22:48:42Z akkad: damn
2014-12-25T22:58:31Z stassats: my arg-checking movement conflicts with store-closure-debug-pointer
2014-12-25T22:59:20Z stassats: since store-closure-debug-pointer has to be done before verifying arg count, because it may be needed when it fails, but it overwrites &more args, since argchecking has to be done before &more, to preserve the arguments
2014-12-25T23:00:27Z stassats: don't see a way past this
2014-12-25T23:01:45Z stassats: i see now, since RAX is not yet overwritten i don't have to query the stack tn in case of invalid-arg-count-error
2014-12-25T23:18:32Z stassats: i think i finally got verify-arg-count before any movements right
2014-12-25T23:20:52Z stassats: and if a special interrupt number is used for invalid arguments, 3 bytes per function can be saved
2014-12-25T23:56:43Z attila_lendvai joined #sbcl
2014-12-25T23:56:43Z attila_lendvai quit (Changing host)
2014-12-25T23:56:43Z attila_lendvai joined #sbcl