00:36:04 -!- hargettp [~anonymous@pool-71-184-188-218.bstnma.east.verizon.net] has quit [Quit: hargettp] 01:14:30 -!- attila_lendvai [~attila_le@89.135.206.215] has quit [Quit: Leaving.] 02:59:27 -!- The_Jon_Smith [~The_Jon_S@ip24-250-13-137.ri.ri.cox.net] has quit [Ping timeout: 265 seconds] 04:07:15 The_Jon_Smith [~The_Jon_S@ip24-250-13-137.ri.ri.cox.net] has joined #sbcl 05:44:58 -!- cmm [~cmm@109.66.199.216] has quit [Ping timeout: 252 seconds] 05:45:51 cmm [~cmm@bzq-109-66-199-216.red.bezeqint.net] has joined #sbcl 05:46:39 mega1 [~quassel@catv4E5CABA2.pool.t-online.hu] has joined #sbcl 05:50:19 -!- cmm [~cmm@bzq-109-66-199-216.red.bezeqint.net] has quit [Ping timeout: 240 seconds] 05:50:52 cmm [~cmm@109.67.202.201] has joined #sbcl 05:55:51 tcr [~tcr@cpc1-bour2-0-0-cust414.15-1.cable.virginmedia.com] has joined #sbcl 05:56:31 tcr1 [~tcr@cpc5-bour5-2-0-cust340.15-1.cable.virginmedia.com] has joined #sbcl 06:00:58 -!- tcr [~tcr@cpc1-bour2-0-0-cust414.15-1.cable.virginmedia.com] has quit [Ping timeout: 272 seconds] 06:03:48 -!- tcr1 [~tcr@cpc5-bour5-2-0-cust340.15-1.cable.virginmedia.com] has quit [Quit: Leaving.] 07:05:38 cmm- [~cmm@bzq-109-64-205-76.red.bezeqint.net] has joined #sbcl 07:08:30 -!- cmm [~cmm@109.67.202.201] has quit [Ping timeout: 240 seconds] 07:18:54 attila_lendvai [~attila_le@89.135.206.215] has joined #sbcl 07:20:19 stassats [~stassats@wikipedia/stassats] has joined #sbcl 07:21:48 -!- Krystof [~csr21@78.146.225.217] has quit [Ping timeout: 245 seconds] 08:07:22 Krystof [~csr21@84-51-132-95.christ977.adsl.metronet.co.uk] has joined #sbcl 08:07:23 -!- ChanServ has set mode +o Krystof 08:35:21 tcr [~tcr@cpc5-bour5-2-0-cust340.15-1.cable.virginmedia.com] has joined #sbcl 08:45:11 -!- Krystof [~csr21@84-51-132-95.christ977.adsl.metronet.co.uk] has quit [Ping timeout: 265 seconds] 08:46:47 nikodemus [~nikodemus@cs181199216.pp.htv.fi] has joined #sbcl 08:46:47 -!- ChanServ has set mode +o nikodemus 09:08:29 good morning 09:12:05 hi, I wonder if I should submit my contrib to ecl right away (with an appropriate 'experimental' warning) 09:21:30 -!- cmatei [~cmatei@95.76.24.124] has quit [Read error: No route to host] 09:29:51 nikodemus: here is the thing, separated into its own repository: http://github.com/angavrilov/cl-simd 09:54:56 excellent! 09:55:03 looking at it now 09:57:19 Of course, it requires all the patches to the core by pkhuong and me 10:02:19 angavrilov: it would be better to document the SSE-PACK type using @anchor and @deftp, so it gets indexed 10:02:31 nikodemus pasted "something like this, example from the docs we generate for SB-MD5" at http://paste.lisp.org/display/114944 10:02:56 angavrilov: but that's a nit 10:03:34 overall i like it. i can list other things i would do differently if it were my show, but it isn't :) 10:04:19 similarly for functions, etc, of course 10:10:46 Krystof [~csr21@cpc2-dals3-0-0-cust1263.hari.cable.virginmedia.com] has joined #sbcl 10:10:46 -!- ChanServ has set mode +o Krystof 11:03:26 -!- stassats [~stassats@wikipedia/stassats] has quit [Read error: Operation timed out] 11:08:05 What is that @phantom{concurrency:} for in the texinfo macros? 11:08:57 to get the pdf output with nice outdents 11:11:49 How does it work? 11:14:31 phantom produces a space the size of its argument, without actually printing anything. llap produces output right-aligned to the current position, without advancing the position 11:14:39 the net effect is that the colon is always in the same horizontal position 11:15:05 (and the sb-) pokes out into the left margin, but that was the best I could do 11:22:35 -!- attila_lendvai [~attila_le@89.135.206.215] has quit [Quit: Leaving.] 11:31:22 nikodemus: do you think .43 and .50 are easily fixable, or should I just revert them for now? 11:37:08 rbarraud [~rbarraud@118-93-95-133.dsl.dyn.ihug.co.nz] has joined #sbcl 12:02:48 -!- rbarraud [~rbarraud@118-93-95-133.dsl.dyn.ihug.co.nz] has quit [Ping timeout: 240 seconds] 12:11:21 -!- nikodemus [~nikodemus@cs181199216.pp.htv.fi] has quit [Quit: This computer has gone to sleep] 12:35:11 stassats [~stassats@wikipedia/stassats] has joined #sbcl 13:26:34 -!- tcr [~tcr@cpc5-bour5-2-0-cust340.15-1.cable.virginmedia.com] has quit [Quit: Leaving.] 13:28:28 -!- Krystof [~csr21@cpc2-dals3-0-0-cust1263.hari.cable.virginmedia.com] has quit [Ping timeout: 245 seconds] 13:53:51 tcr [~tcr@cpc5-bour5-2-0-cust340.15-1.cable.virginmedia.com] has joined #sbcl 13:55:01 Will SBCL inline code, essentially skipping method dispatch, on (gf (the (foo))) ? 13:57:25 tcr: no 13:57:42 bummer 13:58:08 tcr: can't, with method redefinitions and combinations. 13:58:58 you could optimistically do it for standard combination, but that would require more cleverness than sbcl has currently 14:05:31 iirc krystof has a proof of concept patch for that somewhere 14:07:01 has anyone looked at TomP's darwin build problem? 14:08:15 it looks like he copied build trees without cleaning 14:22:55 attila_lendvai [~attila_le@catv-89-133-171-82.catv.broadband.hu] has joined #sbcl 14:31:01 nyef [~nyef@pool-70-109-134-127.cncdnh.east.myfairpoint.net] has joined #sbcl 14:31:17 G'morning all. 14:34:57 hi nyef 14:35:58 *nyef* suspects that today will be spent bisecting to find bugs. 14:45:08 uh why is the value in sb-ext:defglobal tried to be evaluated at compile time, too? 14:46:10 It's probably going to turn out to be a "nobody thought it was a bad idea, and the other def* variable things do it..." thing. 14:46:48 nope defvar/parameter don't 14:47:04 Really? Hunh. 14:47:18 Why should they? 14:47:35 Dunno. Clearly not thinking straight yet. 14:48:58 ah, ok. thanks froydnj. 14:49:21 I was afraid it was a non-sb-thread problem, but, no, that builds fine too. 14:49:33 slyrus_: Dare I ask? 14:50:02 nyef: no, just a probably bogus bug report on lp 14:50:10 Fair enough. 14:50:14 lp646398 14:50:32 lp 646398 14:50:33 https://bugs.launchpad.net/bugs/646398 14:51:25 Ah, right. That one. 14:51:39 thanks stassats and specbot 14:52:01 I think there may be a new test suite failure on PPC, but haven't verified or tried tracking it down yet. 14:52:13 Probably will this afternoon. 14:54:01 -!- froydnj [~froydnj@gateway.codesourcery.com] has quit [Remote host closed the connection] 14:59:13 nikodemus [~nikodemus@cs181058025.pp.htv.fi] has joined #sbcl 14:59:13 -!- ChanServ has set mode +o nikodemus 14:59:35 Hello nikodemus. 15:04:11 froydnj [~froydnj@gateway.codesourcery.com] has joined #sbcl 15:29:43 -!- ASau [~user@83.69.227.32] has quit [Ping timeout: 245 seconds] 15:29:55 hi 15:31:08 Would you happen to know anything about the non-unicode build failures, or should I continue trying to figure out how to fix them? 15:31:20 haven't looked at them, no 15:31:24 Fair enough. 15:31:32 Guess that's my project for this afternoon. 15:32:14 jsnell: .43 is taken care of now, and I don't experience any issues with :timer :stress using that tree -- but i didn't verify that it happened on that linux box before i fixed the first regression 15:32:50 rebuilding without the fix now to see if i can reproduce the :timer :stress thing 15:34:09 I saw the :timer :stress thing with .43 / .45 reverted, so I don't think that's the reason 15:34:45 did it happen every time? 15:35:01 at least several times in a row 15:35:10 hm 15:35:55 if i can't reproduce it, i'll #!+darwin the .50 15:36:49 ho, ho. doesn't happen on my server 15:37:11 nikodemus: What's about always-bound that requires the value known at compile-time? 15:37:13 so I guess I'll need to have a look at it tonight 15:37:54 tcr: the value doesn't have to be known, but it has to be initialized to _something_ 15:38:46 and keeping the value consistent seemed the cleanest option 15:39:38 if you don't need that, you can do (defglobal **foo** nil) (setf **foo** (my-load-time-computation)) 15:40:34 But why does it need to be initialized to something at compile time? 15:40:41 tcr: consider using an always-bound variable during compilation 15:41:08 (at least, I'm guessing that's why nikodemus did what he did) 15:41:19 Hm ok 15:41:32 (declaim (special )) has compile-time effects, too, I see :-) 15:49:17 froydnj has it 15:49:37 just clarified it email for the eavesdroppers on sbcl-devel 15:50:30 jsnell: i can't reproduce the :timer :stress 15:51:05 so sticking a #!+darwin on it 15:51:46 (the nanosleep post-test, that is) 15:52:49 please don't 15:53:03 I'll investigate why it's failing for me 15:53:30 but sprinkling reader conditionals around for things we don't understand would be bad 15:54:00 ok 15:55:00 nikodemus: why is there the os-provides-poll feature ? 15:55:23 every sane Unix system has poll() 15:55:25 even AIX 15:55:53 probably because darwin's poll is braindead wrt stdin 15:57:28 and /dev/zero, etc 15:58:43 rmarynch [~roman@88.135.194.233] has joined #sbcl 15:58:49 but make-config.sh doesn't seem to special-case Darwin wrt. poll() 16:01:14 fe[nl]ix: tools-for-build/grovel-features.sh and tools-for-build/os-provides-poll-test.c 16:01:34 anyway, I have a patch for os-simple-poll: http://paste.lisp.org/+2GP3 16:02:33 fe[nl]ix: what's the motivation? 16:02:42 POLLERR needs to be handled, because it's possible that poll() returns 1, but only POLLERR is present in the return 16:03:13 same for select 16:03:55 for example, if the remote host becomes unreachable and the socket gets closed by the kernel 16:05:02 hm, but then UNIX-SIMPLE-POLL should return T 16:05:15 because T means "waiting more won't help" 16:05:45 but makes sense, yeah 16:06:00 I'd expect T to mean that a read/write can be performed 16:06:12 fe[nl]ix: UNIX-SIMPLE-POLL is an internal function 16:07:27 we loop on UNIX-SIMPLE-POLL returning NIL, so it is important that NIL means "nothing yet", and T "something" 16:07:57 it's not that simple 16:08:03 ? 16:08:18 -!- attila_lendvai [~attila_le@catv-89-133-171-82.catv.broadband.hu] has quit [Quit: Leaving.] 16:08:52 if the polling is for :output, then if POLLERR is T it doesn't matter whether POLLOUT is T as well, because you need to close the socket immediately 16:09:36 i don't follow 16:10:02 if polling is for :input, then if POLLIN is T, you can read some bytes, after which if POLLERR is also T, then the socket should be closed 16:10:06 the subsequent write-attempt will fail, error will be signaled, etc -- and the socket is closed 16:12:27 ok, how about http://paste.lisp.org/+2GP3/1 ? 16:12:53 oops, the logand should be logior 16:12:53 stassats` [~stassats@wikipedia/stassats] has joined #sbcl 16:13:12 i was just about to say :) 16:13:51 http://paste.lisp.org/+2GP3/2 16:14:10 looks good to me 16:14:21 do you have a test-case? 16:14:38 not something that can be run locally 16:15:20 but something? 16:15:33 -!- stassats [~stassats@wikipedia/stassats] has quit [Ping timeout: 245 seconds] 16:15:43 the only way to test this locally that comes to my mind would be to use raw sockets to generate fake ICMP packets 16:17:00 fe[nl]ix: a non-local test-case would still be helpful -- even just a verbal description of a scenario where this matters 16:17:45 then i can stick that into the commit message or a comment 16:18:02 you're connected to a server, when it crashes 16:18:50 your kernel will try a TCP retransmission, but the router to which the server was connected will issue a Host Unreachable ICMP 16:21:52 thanks 16:25:19 ASau [~user@83.69.227.32] has joined #sbcl 16:25:45 So, I'm thinking that getting rid of the TLS-allocation junk from the C runtime would be a good idea... 16:31:05 fe[nl]ix: so the select() based version doesn't really need to deal with error-fds after all 16:32:15 why not ? 16:32:19 hm, and possibly POLLPRI should actually not be used there at all, since that's out-of-band, and consumers of fd-streams are only intrested in in-band 16:33:29 because if the count is 1, something happened -- either the event we wanted, or the server crashed 16:36:00 *nyef* suddenly has a possible use for 64-bit unboxed bitfield manipulation... 16:36:29 nikodemus: you're right 16:37:10 actually, i wonder if there's any point in testing the events for poll either 16:37:31 just check for -1/0/1 16:37:53 yepp 16:39:50 -!- rmarynch [~roman@88.135.194.233] has quit [Quit: Leaving] 16:43:05 i wonder what select would do if i passed the same fd-set as the input/output and error-fds... 16:43:10 *nikodemus* resists temptation 16:43:34 nikodemus: are you working on this ? 16:45:21 fe[nl]ix: fsvo working 16:52:47 hm, so, if we have and fd to a socket on server REMOTE, and it goes down, and subsequently we call select() with NULL for write-fds and error-fds but the said fd in read-fds, is select() allowed to return 0 when we time out? 16:53:45 I have no idea 16:54:00 for safety I think we should always pass a non-NULL error-fds to select() 16:54:03 http://paste.lisp.org/+2GP3/3 16:55:26 and fd-set the descriptor we care about there... 16:56:35 http://paste.lisp.org/+2GP3/4 17:15:01 ... Nobody landed wider-fixnums while I wasn't looking, did they? 17:16:34 or a 65-bit port? 17:17:34 I don't think I'd be running a 65-bit port without knowing it, would I? 17:21:15 Krystof [~csr21@78.146.225.217] has joined #sbcl 17:21:15 -!- ChanServ has set mode +o Krystof 17:24:39 i thought it might be a possibility if you were thinking about 64-bit fixnums :) 17:26:06 tsuru [~charlie@c-174-50-217-160.hsd1.tn.comcast.net] has joined #sbcl 17:29:00 -!- ASau [~user@83.69.227.32] has quit [Read error: Connection reset by peer] 17:29:52 error-fds is a misnomer... 17:29:57 unless you're on windows 17:30:16 "exceptfds" doesn't mean error, it means OOB data. 17:31:41 so you really shouldn't be watching it, unless you're planning on doing something with it 17:34:00 i was just starting to wonder how come serve-event users hadn't ever complained about issues with it given its lack of error-fds usage 17:34:36 on windows, it actually means error, because the people writing winsock were just as confused as everyone else 17:34:53 but they were writing the TCP stack instead of just trying to use it. :) 17:43:24 So, I've got something weird. I hacked the %special-unbind ir2-convert method to pass the symbol being unbound to the VOP. If I hack the VOP to make use of that symbol to find the TLS slot, the build fails on the second invocation of DEFCLASS. But if I compare it against the symbol in the binding stack entry, it's right. WTF? 18:07:45 -!- nikodemus [~nikodemus@cs181058025.pp.htv.fi] has quit [Quit: This computer has gone to sleep] 18:08:26 -!- stassats` [~stassats@wikipedia/stassats] has quit [Ping timeout: 264 seconds] 18:27:04 ASau [~user@83.69.227.32] has joined #sbcl 18:28:46 clhs do-all-symbols 18:28:46 http://www.lispworks.com/reference/HyperSpec/Body/m_do_sym.htm 18:39:02 Heh. Random factoid: The first UNBIND that genesis sees is for SB!IMPL::*BACKQUOTE-COUNT*. 18:42:49 kclifton [~kclifton@s198-166-45-245.ab.hsia.telus.net] has joined #sbcl 19:31:47 -!- Blkt [~user@93-33-131-26.ip44.fastwebnet.it] has quit [Ping timeout: 276 seconds] 19:44:07 -!- mbohun [~mbohun@ppp115-156.static.internode.on.net] has quit [Quit: Leaving] 20:03:14 hargettp [~anonymous@pool-71-184-188-218.bstnma.east.verizon.net] has joined #sbcl 21:18:55 Lovely. Something is storing a NIL into the no-tls-value-marker slot. 21:19:39 slyrus__ [~chatzilla@dsl092-019-253.sfo1.dsl.speakeasy.net] has joined #sbcl 21:23:12 Oh, right. Can't :trace-file t when building PCL. :-/ 21:42:18 Hrm. I missed a trick. Instead of just checking for a blown no-tls-value-marker, I should also set it back to the way it should be. 21:42:23 -!- mega1 [~quassel@catv4E5CABA2.pool.t-online.hu] has quit [Ping timeout: 255 seconds] 21:50:48 -!- Krystof [~csr21@78.146.225.217] has quit [Read error: Operation timed out] 22:05:59 Krystof [~csr21@78.146.236.83] has joined #sbcl 22:05:59 -!- ChanServ has set mode +o Krystof 22:13:49 rbarraud [~rbarraud@118-93-95-133.dsl.dyn.ihug.co.nz] has joined #sbcl 23:19:16 ah, brilliant. looks like linux nanosleep can return a slightly higher value in "rem" than was given as input in "req", probably due to rounding to the nearest HZ 23:20:49 -!- tcr [~tcr@cpc5-bour5-2-0-cust340.15-1.cable.virginmedia.com] has quit [Quit: Leaving.] 23:22:32 tcr [~tcr@cpc5-bour5-2-0-cust340.15-1.cable.virginmedia.com] has joined #sbcl 23:22:47 hahaha 23:24:43 -!- foom [~jknight@ita4fw1.itasoftware.com] has quit [*.net *.split] 23:28:52 foom [~jknight@ita4fw1.itasoftware.com] has joined #sbcl 23:31:41 -!- hargettp [~anonymous@pool-71-184-188-218.bstnma.east.verizon.net] has quit [Quit: hargettp] 23:41:17 hargettp [~anonymous@pool-71-184-188-218.bstnma.east.verizon.net] has joined #sbcl 23:42:48 -!- tsuru [~charlie@c-174-50-217-160.hsd1.tn.comcast.net] has quit [Remote host closed the connection] 23:43:07 -!- kclifton [~kclifton@s198-166-45-245.ab.hsia.telus.net] has quit [Quit: kclifton] 23:56:52 -!- The_Jon_Smith [~The_Jon_S@ip24-250-13-137.ri.ri.cox.net] has left #sbcl