00:23:21 -!- kclifton [~kclifton@s198-166-45-245.ab.hsia.telus.net] has quit [Quit: kclifton] 00:29:45 ugh 00:29:59 All the os-provides-* tests fail on OpenBSD now 00:31:54 so, anybody got any more insights into those condition printing memory fault? 00:32:07 Oh, tools-for-build/Makefile includes src/runtime/Config 00:32:54 On BSD-derived systems, that then includes another Config.foo-bsd makefile, which gmake can't find because it's in the wrong directory. 00:46:06 stassats: What? Beyond something having declared a safety so low that it killed type-checking? 00:46:41 yes, beyond that, anyway, i've filled a ticket 00:47:09 I suppose you could either look through the code-path to see what's looking at the CAR of the list or try to get a backtrace from the error. 00:47:57 well, i've already found that it's the result of (apply #'format ... 88) 00:48:17 ... Yeah, that'd do it. Do you know where the APPLY is? 00:48:18 lp 705690 00:48:18 https://bugs.launchpad.net/bugs/705690 00:48:19 -!- slyrus_ [~chatzilla@173-228-44-88.dsl.static.sonic.net] has quit [Ping timeout: 240 seconds] 00:48:53 ... The function SB-SYS:FIND-DYNAMIC-FOREIGN-SYMBOL-ADDRESS is undefined? 00:49:40 1.0.45.9 x86-64/linux fails to build asdf-install, sb-bsd-sockets, sb-posix and sb-simple-streams. 00:50:22 is this related to recent -ldl issues? 00:50:30 Might be. 00:50:55 But this particular host is running debian stable, and hasn't been updated in months. 00:51:00 Does the local-target-features lack all the os-* features? 00:51:15 nyef: well, maybe the fix for new systems broke it on old 00:51:18 Good call. That is, indeed, the case. 00:51:33 Yea, the tools-for-build makefile is broken. 00:52:42 It must be for a different reason than I'm seeing though, since the linux Config.foo-linux makefiles don't include any other makefiles. 00:53:27 stassats: How quickly can you do a full build? 00:53:47 9-10 minutes 00:54:38 That's about three times faster than me. Can you try wrapping a (locally (declare (optimize (safety 2))) ...) around the definition of simple-condition-printer in src/code/condition.lisp and see if it "fixes" the bug? (safety 1) might work as well. 00:55:13 wrapping around it? 00:55:30 Yeah. Have the LOCALLY at toplevel and the DEFUN in its body. 00:55:36 and not putting the declaration inside? 00:55:38 why is that? 00:56:32 Because my mental model of where declarations take effect is a little fuzzy, and having the declaration at toplevel guarantees that it'll affect the entire function. 00:56:48 I don't understand why the includes in the makefile were changed, they should have worked fine before. 00:57:06 ./clean.sh Makefile:10: ../src/runtime/genesis/Makefile.features: No such file or directory 00:57:15 Yes, the broken makefile again. 00:57:18 joshe: is this related too what you're talking? 00:57:41 slyrus: why did you change the include lines in tools-for-build/Makefile ? 00:58:02 Back out to .5 or so and try again? 00:58:25 Or just try this: http://www.elsasser.org/misc/sbcl-tfb-makefile.diff 00:59:16 i wonder how fast is it to build sbcl on recently released core iX 01:00:05 Perhaps he only wanted to move $(OS_LIBS) to LDLIBS and the include bits snuck in by mistake. 01:12:50 nyef: locally (declare (optimize (safety 2))) doesn't help 01:13:08 Hrm. 01:14:26 it doesn't seem to be a safety issue, because i can see in disassembly that it checks other things 01:15:16 Yeah, the structure slot is type-restricted, isn't it? 01:15:28 Yeah, the slot is :type list. 01:15:46 well, when i compile it with C-c C-c, it gets a type-check 01:17:04 Hrm. That means either a hunt for the actual policy controls of interest, or trying to figure out why it's creating a bogus instance in the first place. 01:17:57 FWIW, an sbcl build with the above patch finishes fine, with all the contribs 01:18:03 slot type checking on classes is done on safety 2 or 3 01:18:26 i don't know whether this is true for condition-classes 01:18:42 I'm not sure that it doesn't re-break whatever slyrus was trying to fix though. 02:44:45 -!- hargettp [~hargettp@96.237.131.147] has quit [Quit: Leaving...] 03:02:04 -!- homie [~levgue@xdsl-78-35-166-152.netcologne.de] has quit [Read error: Connection reset by peer] 03:03:01 homie [~levgue@xdsl-78-35-141-172.netcologne.de] has joined #sbcl 03:06:26 cmm- [~cmm@bzq-79-177-204-134.red.bezeqint.net] has joined #sbcl 03:10:05 -!- cmm [~cmm@bzq-79-178-206-31.red.bezeqint.net] has quit [Ping timeout: 276 seconds] 04:15:31 -!- nyef [~nyef@pool-70-20-57-204.man.east.myfairpoint.net] has quit [Quit: G'night all.] 04:18:20 joshe: did you get things working/ 04:18:24 ? 04:18:43 broken shift key here it seems... 05:02:49 Yea, did you see that patch? 05:03:11 Reverting the change to the two include lines in tools-for-build/Makefile makes things work for me. 05:03:49 yeah, but that patch was there for a reason 05:04:10 my take on the old version is that the include lines are silently failing 05:04:45 When I grepped around it looked like gmake was always invoked with the right include path. 05:04:48 the problem is that the genesis/Makefile.features may not exist yet, causing things to fail 05:05:11 When I made gmake noisy, I could see that it was picking up flags from the src/runtime makefiles. 05:05:21 Yes, that was one problem. 05:05:23 but one needs the ../src/runtime/Config to get the right linker options for recent-ish linux builds 05:06:00 so I think deleting the Makefile.features is probably the right thing to do and we need to make sure that Config gets properly included 05:06:07 it works for me here after a clean.sh 05:06:15 I need to reboot, but I'll be back in a few minutes 05:06:36 hm 05:07:03 think about it while I'm gone... :) 05:07:21 Actually, if gmake always sets up the include path correctly then I shouldn't see failures with nested inlcudes either way. 05:07:29 -!- slyrus [~chatzilla@adsl-75-36-217-249.dsl.pltn13.sbcglobal.net] has quit [Remote host closed the connection] 05:21:20 eh, I'm too tired to figure this out tonight 05:26:46 slyrus [~chatzilla@adsl-75-36-217-249.dsl.pltn13.sbcglobal.net] has joined #sbcl 05:27:38 joshe: OK, I'm willing to concede that both of those include lines should be -include 05:27:52 but the path is (was) wrong, at least for darwin and linux 05:28:03 perhaps openbsd's make does things differently 05:28:15 It's gnu make everywhere 05:28:36 The makefiles don't work with normal make 05:28:51 ok, then they should be the same everywhere 05:29:04 as they used to be, they were wrong, as I understand it 05:29:16 Why is that? 05:29:38 And from what I can see, it's invoked with an include path that should have worked everywhere, and appeared to work for me 05:30:31 But there's something funny there that I can't understand, and I'm tired, so I'll go to bed and try to figure it out tomorrow 05:30:48 well, it wasn't working for me until I added the ../ 05:32:06 if you believe that the -I is working, the include should just be genesis/Makefile.features and Config 05:32:30 at least based on the -I ../src/runtime from grovel-features.sh 06:34:50 -!- stassats [~stassats@wikipedia/stassats] has quit [Ping timeout: 276 seconds] 07:25:52 flip214 [~marek@2001:858:107:1:baac:6fff:fe6b:9183] has joined #sbcl 07:25:52 -!- flip214 [~marek@2001:858:107:1:baac:6fff:fe6b:9183] has quit [Changing host] 07:25:52 flip214 [~marek@unaffiliated/flip214] has joined #sbcl 07:38:31 |3b|``` [foobar@cpe-72-179-19-4.austin.res.rr.com] has joined #sbcl 07:38:36 -!- |3b|`` [foobar@cpe-72-179-19-4.austin.res.rr.com] has quit [Remote host closed the connection] 08:51:51 -!- ASau [~user@93-80-210-26.broadband.corbina.ru] has quit [Ping timeout: 240 seconds] 09:06:12 -!- homie [~levgue@xdsl-78-35-141-172.netcologne.de] has quit [Quit: ERC Version 5.3 (IRC client for Emacs)] 09:10:28 homie [~levgue@xdsl-78-35-141-172.netcologne.de] has joined #sbcl 09:43:17 tcr1 [~tcr@213.55.131.17] has joined #sbcl 10:14:38 hargettp [~hargettp@96.237.131.147] has joined #sbcl 10:22:44 mega1 [~quassel@catv4E5CABA2.pool.t-online.hu] has joined #sbcl 10:43:46 -!- tcr1 [~tcr@213.55.131.17] has quit [Read error: Connection reset by peer] 10:50:35 tcr1 [~tcr@217-162-130-31.dynamic.hispeed.ch] has joined #sbcl 11:00:59 tcr2 [~tcr@217-162-130-31.dynamic.hispeed.ch] has joined #sbcl 11:01:03 -!- tcr1 [~tcr@217-162-130-31.dynamic.hispeed.ch] has quit [Ping timeout: 240 seconds] 11:28:39 -!- tcr2 [~tcr@217-162-130-31.dynamic.hispeed.ch] has quit [Ping timeout: 240 seconds] 12:02:10 tcr1 [~tcr@80-218-246-66.dclient.hispeed.ch] has joined #sbcl 12:02:32 stassats [~stassats@wikipedia/stassats] has joined #sbcl 12:08:50 -!- hargettp [~hargettp@96.237.131.147] has quit [Quit: Leaving...] 12:11:09 hargettp [~hargettp@96.237.131.147] has joined #sbcl 13:18:22 attila_lendvai [~attila_le@catv-80-98-24-21.catv.broadband.hu] has joined #sbcl 13:32:46 -!- hargettp [~hargettp@96.237.131.147] has quit [Quit: Leaving...] 13:34:04 hargettp [~hargettp@96.237.131.147] has joined #sbcl 13:55:43 -!- flip214 [~marek@unaffiliated/flip214] has quit [Remote host closed the connection] 13:56:37 -!- tcr1 [~tcr@80-218-246-66.dclient.hispeed.ch] has quit [Ping timeout: 255 seconds] 13:56:44 tcr1 [~tcr@80-218-246-66.dclient.hispeed.ch] has joined #sbcl 14:17:02 -!- Krystof [~csr21@csrhodes.plus.com] has quit [Quit: Leaving] 14:38:42 nyef [~nyef@pool-70-20-57-204.man.east.myfairpoint.net] has joined #sbcl 14:39:11 G'morning all. 14:39:42 slyrus: "fsbcl-or"? 15:13:06 Okay, clean builds on 2/3 of my SBCL targets. I think that's good enough. 15:13:12 pkhuong pasted "Portable, generic opaque identity" at http://paste.lisp.org/display/118882 15:15:09 Shiny! How's it come in handy? 15:16:21 to manually split live ranges, mostly 15:16:49 I should add an optional type argument 15:17:10 it's also useful to foil type inference sometimes. 15:18:12 Ahh. 15:19:13 pkhuong: why would one want to do that ? 15:19:36 split live ranges? 15:19:52 foil type inference 15:21:02 it's (again) mostly related to representation selection 15:21:02 ... When adding a single-float and a fixnum, and both values are boxed, the compiler just does a generic-+ ? 15:21:51 nyef: did I mess something up? 15:22:02 slyrus: Have a look at version.lisp-expr. 15:22:25 It's not really worth a separate commit, but it's something that should probably get fixed down the line. 15:22:27 fe[nl]ix: e.g. sometimes you really want a machine word representation, even though the values fit in fixnums 15:22:40 oops! wrong window apparently. 15:22:47 sorry 'bout that 15:23:12 well, the commit count has been a bit light lately. this well help. 15:23:49 pkhuong: It's too bad we can't do cost calculations over the entire function as part of template selection, looking for the cheapest representation for LVAR TNs, huh? 15:24:52 nyef: I've thought about that 15:24:54 antgreen [~user@CPE00222d6c4710-CM00222d6c470d.cpe.net.cable.rogers.com] has joined #sbcl 15:25:05 I do have the hack planned for using unboxed value storage for d-x closures, even beyond eliminating the value-cells, though. 15:25:33 it's pretty much a quadratic binary programming problem; very hard to solve exactly, but not too hard to approximate in practice 15:25:42 homie` [~levgue@xdsl-78-35-137-186.netcologne.de] has joined #sbcl 15:26:04 problem is, even that won't be enough to compensate for the fact that we associate all that information to variables 15:26:20 ... and, even worse, sometimes (often) *merge* variables. 15:26:26 Not the LTNs, though! 15:26:49 LTNs are only for temporaries, no? 15:26:54 Right. 15:28:53 -!- homie [~levgue@xdsl-78-35-141-172.netcologne.de] has quit [Ping timeout: 264 seconds] 15:29:03 -!- homie` [~levgue@xdsl-78-35-137-186.netcologne.de] has quit [Client Quit] 15:35:03 -!- stassats [~stassats@wikipedia/stassats] has quit [Ping timeout: 240 seconds] 15:39:31 -!- hargettp [~hargettp@96.237.131.147] has quit [Quit: Leaving...] 15:41:04 nyef: those tools-for-build/Makefile changes look OK to you? 15:41:35 In .10? 15:42:05 yup 15:42:45 -!- tcr1 is now known as tcr 15:43:40 It's basically putting the two include lines back to the way they were prior to .9, right? 15:44:07 slyrus: I thought it wasn't working on linux with the full path? 15:45:06 It (.10) builds on my x86-64 and ppc systems now, so I'm happy. 15:45:42 Yea, my nightly openbsd amd64 build went fine last night with .10 15:46:03 yes, but OS_LIBS needed to be added to LDLIBS not LDFLAGS 15:46:35 ah 15:46:59 The default implicit rule puts LDFLAGS at the beginning and LDLIBS at the end? 15:48:26 I think that's correct 15:48:50 Krystof [~csr21@csrhodes.plus.com] has joined #sbcl 15:48:50 -!- ChanServ has set mode +o Krystof 15:50:26 uh oh, the moderator is here. we had better behave. 15:50:37 have you been breaking everything in my absence? 15:50:59 *Krystof* reads the latest sbcl-commits message 15:51:01 *Krystof* giggles 15:51:12 it doesn't count if it ends up being fine by the release ;) 15:51:23 yeah, what is that release manager doing? 15:51:42 Are we going into code freeze any time soon? 15:52:19 dunno. I say break HEAD some more, if it's not currently already broken 15:52:30 we haven't had any exciting new features in aaaages 15:52:56 I have another change to d-x closures that could go in, or I can hold off on it. It's something like two lines of code change, eight lines of comment, and twenty lines of commit message. 15:53:54 I think we should resist the temptation to hold off on things for too long 15:54:16 look at what happens to all the neat hacks if they don't get put in -- where's my sb-regpair!? 15:54:19 Does that mean I should land the tls-index fixups as well? 15:54:35 unless you think they're really broken (rather than, say, are not sure that they're broken) 15:54:45 otherwise people won't test them properly 15:55:34 No, this is more along the lines of further optimization, not anything that affects correctness. 15:56:10 -!- slyrus [~chatzilla@adsl-75-36-217-249.dsl.pltn13.sbcglobal.net] has quit [Ping timeout: 240 seconds] 16:12:16 Land'em! further optimization can always be done in separate commits. 16:12:44 Just doing a pre-commit test build now. 16:12:59 For the dx-closure patch. 16:13:07 stassats [~stassats@wikipedia/stassats] has joined #sbcl 16:13:25 I should probably be clocking this, but... Meh. 16:23:42 slyrus [~chatzilla@adsl-75-36-217-249.dsl.pltn13.sbcglobal.net] has joined #sbcl 16:24:23 homie [~levgue@xdsl-78-35-137-186.netcologne.de] has joined #sbcl 16:42:29 ... 1.0.45.13? What'd I miss?!? 16:42:53 Ah, more slyrusness. 16:43:28 -!- tcr [~tcr@80-218-246-66.dclient.hispeed.ch] has quit [Quit: Leaving.] 16:43:41 tcr [~tcr@80-218-246-66.dclient.hispeed.ch] has joined #sbcl 16:53:14 -!- tcr [~tcr@80-218-246-66.dclient.hispeed.ch] has quit [Read error: Connection reset by peer] 16:59:45 tcr [~tcr@80-218-246-66.dclient.hispeed.ch] has joined #sbcl 16:59:46 -!- tcr [~tcr@80-218-246-66.dclient.hispeed.ch] has quit [Read error: Connection reset by peer] 16:59:53 tcr [~tcr@80-218-246-66.dclient.hispeed.ch] has joined #sbcl 17:44:25 -!- stassats [~stassats@wikipedia/stassats] has quit [Ping timeout: 246 seconds] 17:53:36 nyef: if you finish that ARM port, perhaps lichtblau can bolt cloak onto in and then you can sell it to the handset mfgrs as the replacement for android 17:53:43 s/onto in/onto it/ 17:55:04 Heh. 17:55:53 The last entry in my local copy of the arm-port-log, dated 2010-Oct-07, basically says "you're doing it wrong". 17:58:53 that's better than "steal code from oracle; change license" I suppose 18:05:58 stassats [~stassats@wikipedia/stassats] has joined #sbcl 18:50:15 -!- slyrus [~chatzilla@adsl-75-36-217-249.dsl.pltn13.sbcglobal.net] has quit [Ping timeout: 240 seconds] 19:06:26 Heh. They only allegedly stole like 100LOC. 19:39:27 tcr1 [~tcr@80-218-246-66.dclient.hispeed.ch] has joined #sbcl 19:40:11 -!- tcr [~tcr@80-218-246-66.dclient.hispeed.ch] has quit [Read error: Connection reset by peer] 20:01:33 -!- antgreen [~user@CPE00222d6c4710-CM00222d6c470d.cpe.net.cable.rogers.com] has quit [Read error: Operation timed out] 20:02:13 antgreen [~user@CPE00222d6c4710-CM00222d6c470d.cpe.net.cable.rogers.com] has joined #sbcl 20:04:32 -!- homie [~levgue@xdsl-78-35-137-186.netcologne.de] has quit [Quit: ERC Version 5.3 (IRC client for Emacs)] 20:07:56 homie [~levgue@xdsl-78-35-137-186.netcologne.de] has joined #sbcl 20:11:15 rmarynch [~roman@88.135.194.233] has joined #sbcl 20:32:51 Good evening 20:33:42 Could somebody please review (and commit?) this small LOOP patch: https://bugs.launchpad.net/sbcl/+bug/645534 20:34:12 I have other patches for LOOP, but I would like them to go one by one 20:39:34 I'll add it to my queue. 20:42:05 nyef: thanks. It is not so easy to find a committer for a patch these days :) 20:58:45 -!- rmarynch [~roman@88.135.194.233] has quit [Quit: Leaving] 21:36:06 slyrus [~chatzilla@173-228-44-88.dsl.static.sonic.net] has joined #sbcl 21:37:09 is it possible to compile a x86 binary from a x86-64 sbcl? 21:39:28 derrida: Yes. 21:39:48 cross-compilation is fairly straightforward, and if you have a multilib machine, well... 22:25:41 -!- antgreen [~user@CPE00222d6c4710-CM00222d6c470d.cpe.net.cable.rogers.com] has quit [Read error: Operation timed out] 22:27:00 antgreen [~user@CPE00222d6c4710-CM00222d6c470d.cpe.net.cable.rogers.com] has joined #sbcl 22:39:31 -!- mega1 [~quassel@catv4E5CABA2.pool.t-online.hu] has quit [Ping timeout: 255 seconds] 22:49:06 -!- stassats [~stassats@wikipedia/stassats] has quit [Ping timeout: 240 seconds] 23:39:08 -!- tsuru [~charlie@adsl-179-29-10.bna.bellsouth.net] has quit [Remote host closed the connection]