00:45:53 -!- cmm [~cmm@bzq-79-180-138-123.red.bezeqint.net] has quit [Read error: Connection reset by peer] 02:17:34 attila_lendvai [~attila_le@92.47.250.52] has joined #sbcl 02:17:34 -!- attila_lendvai [~attila_le@92.47.250.52] has quit [Changing host] 02:17:34 attila_lendvai [~attila_le@unaffiliated/attila-lendvai/x-3126965] has joined #sbcl 02:32:00 cmm [~cmm@bzq-79-179-155-88.red.bezeqint.net] has joined #sbcl 02:39:13 -!- christoph4 [~christoph@ppp-188-174-128-100.dynamic.mnet-online.de] has quit [Ping timeout: 246 seconds] 02:47:43 echo-area [~user@182.92.247.2] has joined #sbcl 02:50:46 -!- Bike [~Glossina@75-175-78-112.ptld.qwest.net] has quit [Ping timeout: 246 seconds] 02:51:58 Bike [~Glossina@75-175-78-112.ptld.qwest.net] has joined #sbcl 02:52:43 christoph4 [~christoph@ppp-188-174-116-180.dynamic.mnet-online.de] has joined #sbcl 02:56:06 -!- Quadrescence [~quad@unaffiliated/quadrescence] has quit [Quit: This computer has gone to sleep] 04:01:00 yacks [~py@180.151.36.168] has joined #sbcl 04:02:21 -!- echo-area [~user@182.92.247.2] has quit [Remote host closed the connection] 04:03:11 echo-area [~user@182.92.247.2] has joined #sbcl 04:11:16 -!- Bike [~Glossina@75-175-78-112.ptld.qwest.net] has quit [Ping timeout: 246 seconds] 04:12:52 Bike [~Glossina@75-175-78-112.ptld.qwest.net] has joined #sbcl 04:25:07 -!- yacks [~py@180.151.36.168] has quit [Quit: Leaving] 04:51:03 -!- attila_lendvai [~attila_le@unaffiliated/attila-lendvai/x-3126965] has quit [Remote host closed the connection] 04:51:53 attila_lendvai [~attila_le@92.47.250.52] has joined #sbcl 04:51:54 -!- attila_lendvai [~attila_le@92.47.250.52] has quit [Changing host] 04:51:54 attila_lendvai [~attila_le@unaffiliated/attila-lendvai/x-3126965] has joined #sbcl 05:00:20 I just had a compiler crash while trying to build the latest SBCL from git 05:03:07 Here's the details: http://paste.lisp.org/display/137330 05:36:05 attila_lendvai1 [~attila_le@92.47.250.52] has joined #sbcl 05:36:05 -!- attila_lendvai [~attila_le@unaffiliated/attila-lendvai/x-3126965] has quit [Disconnected by services] 05:36:05 -!- attila_lendvai1 [~attila_le@92.47.250.52] has quit [Changing host] 05:36:05 attila_lendvai1 [~attila_le@unaffiliated/attila-lendvai/x-3126965] has joined #sbcl 05:39:57 -!- attila_lendvai1 [~attila_le@unaffiliated/attila-lendvai/x-3126965] has quit [Client Quit] 05:40:01 attila_lendvai [~attila_le@92.47.250.52] has joined #sbcl 05:40:01 -!- attila_lendvai [~attila_le@92.47.250.52] has quit [Changing host] 05:40:01 attila_lendvai [~attila_le@unaffiliated/attila-lendvai/x-3126965] has joined #sbcl 05:44:47 edgar-rft [~GOD@HSI-KBW-149-172-63-75.hsi13.kabel-badenwuerttemberg.de] has joined #sbcl 05:51:31 OK, it's not related to the SBCL I was using for building. I tried to rebuild by bootstrapping from ABCL instead, and I got the same behaviour. 06:23:42 prxq [~mommer@mnhm-590c19f0.pool.mediaWays.net] has joined #sbcl 06:35:08 loke_: can you type "print $1" at the ldb prompt? 06:38:12 also, what's your platform? 06:38:53 -!- Bike [~Glossina@75-175-78-112.ptld.qwest.net] has quit [Ping timeout: 256 seconds] 06:41:56 Bike_ [~Glossina@75-175-78-112.ptld.qwest.net] has joined #sbcl 06:42:32 Krystof: amd64 06:42:46 I exited the ldb. I have to do it again. Wait 5 mins or so 06:44:54 you can restart the ldb with src/runtime/sbcl --core output/cold-sbcl.core 06:45:03 what OS? 06:45:28 also what set of optional features? 06:49:01 (because needless to say this failure in cold-init does not happen to me, on amd64/linux) 06:52:33 yacks [~py@180.151.36.168] has joined #sbcl 07:13:18 ok got it 07:13:22 the OS is Linux Arch 07:13:44 no optional features. I just run the default make.sh 07:13:49 -!- Bike_ [~Glossina@75-175-78-112.ptld.qwest.net] has quit [Ping timeout: 245 seconds] 07:14:26 http://paste.lisp.org/display/137331 07:14:31 That's the output from print $1 07:14:49 SETF INFDO 07:14:54 _really_? 07:14:57 Yeah. 07:14:59 What is that? 07:15:13 well it's probably trying to be (SETF INFO) 07:15:30 Are you saying that I may have a typo in the files? 07:15:52 I suppose that could have happened if I Meta-.'ed through the SBCL source 07:15:54 let me check 07:16:26 find . -type f | xargs grep -i setf.*\(infdo 07:16:45 hang on. FOund it, and yep. You're right 07:16:48 the file is modified 07:17:09 yep. thanks 07:17:15 That's it 07:17:34 *phew* 07:18:26 It must have happened when I was reading (and trying to understand) the SBCL source 07:19:10 now you understand a bit more :-) 07:19:15 I think I'll update my emacs config to always set files in certain directories as read-only by default 07:19:28 You know if there 07:19:43 You know if there's already a feature for this, or should I write it myself? 07:21:23 I think you could probably use dir-local variables and buffer-read-only 07:22:44 dir-local variables? I had no idea of those 07:22:49 :-) 07:22:54 Thanks 07:23:10 yay. sbcl built 07:23:13 thanks again 08:27:54 rudi_ [~rudi@1x-193-157-196-114.uio.no] has joined #sbcl 08:28:15 -!- rudi_ [~rudi@1x-193-157-196-114.uio.no] has quit [Client Quit] 08:28:36 rudi_ [~rudi@1x-193-157-196-114.uio.no] has joined #sbcl 08:29:26 -!- rudi_ is now known as rudi 08:54:53 -!- jdz [~jdz@85.254.212.34] has quit [Quit: Byebye.] 08:58:51 jdz [~jdz@85.254.212.34] has joined #sbcl 09:16:24 leoc [~leoc.git@p5480A412.dip0.t-ipconnect.de] has joined #sbcl 09:45:18 scymtym [~user@2001:638:504:2093:226:b9ff:fe7d:3e1f] has joined #sbcl 09:47:31 -!- yacks [~py@180.151.36.168] has quit [Quit: Leaving] 09:58:40 *attila_lendvai* recompiles sbcl and tries to extract a test-case for Krystof 10:04:54 -!- scymtym [~user@2001:638:504:2093:226:b9ff:fe7d:3e1f] has quit [Remote host closed the connection] 10:06:04 scymtym [~user@2001:638:504:2093:226:b9ff:fe7d:3e1f] has joined #sbcl 10:20:57 thank you attila_lendvai 10:23:16 Krystof: any idea what that typo-like thing is in the warnings? they seem to be reproducible here: ";;; warning: wnused declarations found in form:" 10:25:32 that really looks like single-bit error 10:25:41 I mean I don't want to blame bad ram, but 10:26:56 hm, actually, that's not a warning in SBCL sources, is it? 10:39:44 Krystof: no, they are warning on code in cl-l10n. I've seen those errors twice already, once without your latest patch, and now with it sbcl recompiled, fasl's purged 10:40:13 the full warning looks like this: 10:40:15 they're warning on code in cl-l10n, but where is the warning itself defined? 10:40:15 ;;; warning: wnused declarations found in form: (IGNORABLE DATE), (IGNORABLE DAY-OF-WEEK), (TYPE NON-NEGATIVE-FIXNUM DAY-OF-WEEK). 10:40:30 ahh, good question! 10:40:38 that is not a warning that comes from sbcl code 10:41:02 *attila_lendvai* grepped for it in the sbcl sources and associate to the character bugs... typical cognitive error... 10:41:17 at least, I don't think it is a warning in sbcl code 10:42:08 yes, it's a typo in metabang bind 10:42:15 sorry for the false alarm! 11:29:20 -!- echo-area [~user@182.92.247.2] has quit [Remote host closed the connection] 12:27:38 foreignFunction [~niksaak@ip-4761.sunline.net.ua] has joined #sbcl 12:33:13 davazp [~user@92.251.217.119.threembb.ie] has joined #sbcl 13:18:57 wbooze [~wbooze@xdsl-78-35-189-86.netcologne.de] has joined #sbcl 14:19:38 nyef [~nyef@pool-70-109-132-87.cncdnh.east.myfairpoint.net] has joined #sbcl 14:19:57 G'morning all. 14:27:25 I'm in a conference with less than ideal connectivity, but I'll be bisecting and thinking on the parser bug 14:34:47 -!- rudi [~rudi@1x-193-157-196-114.uio.no] has quit [Quit: rudi] 14:48:01 -!- davazp [~user@92.251.217.119.threembb.ie] has quit [Ping timeout: 248 seconds] 14:56:14 -!- attila_lendvai [~attila_le@unaffiliated/attila-lendvai/x-3126965] has quit [Quit: Leaving.] 15:15:30 Quadrescence [~quad@unaffiliated/quadrescence] has joined #sbcl 15:32:35 oh, an easy one. Some problem in the new eq vops. 15:32:53 (if I bisected this right) 15:33:26 odd that it shows up for (values) but not for nil 15:52:36 I'm going to have access to my PPC and SPARC again on Wednesday, and I'm working on getting a semi-usable MIPS system, though I shudder to think how outdated the 2.4 series kernel is, and it's not upgradable from there due to gcc dropping support for the specific CPU involved some time back. 16:00:05 yacks [~py@180.151.36.168] has joined #sbcl 16:09:25 -!- Quadrescence [~quad@unaffiliated/quadrescence] has quit [Read error: Connection reset by peer] 16:10:00 Quadrescence [~quad@unaffiliated/quadrescence] has joined #sbcl 16:12:48 nyef: clang (hah)? 16:13:19 On a PlayStation 2? Hardly seems worth the attempt. 16:13:36 Especially for trying to build a new Linux kernel. 16:13:48 (Which, in turn, would likely require considerable patching.) 16:14:17 there's a patchset floating around. 16:15:52 Perhaps later. My project list only seems to grow these days. (-: 16:22:25 yeah. disabling the new EQ VOPs in pred.lisp fixes this. What load-if magic do I not understand now? (: 16:22:53 Dare I ask? 16:23:37 I thought that :load-if was fairly straightforward, but there could easily be aspects that I don't understand. 16:24:24 I don't know. It's simply the only thing I can think of so far 16:24:52 No, no... I'm missing just about all of the context for your bug-hunt right now. 16:33:59 Bike [~Glossina@75-175-78-112.ptld.qwest.net] has joined #sbcl 17:19:46 -!- yacks [~py@180.151.36.168] has quit [Quit: Leaving] 17:34:39 davazp [~user@178.167.175.194.threembb.ie] has joined #sbcl 18:01:28 -!- davazp [~user@178.167.175.194.threembb.ie] has quit [Read error: Connection reset by peer] 18:37:45 slyrus [~chatzilla@adsl-99-183-240-185.dsl.pltn13.sbcglobal.net] has joined #sbcl 18:43:03 drmeister [~drmeister@pool-71-185-168-200.phlapa.fios.verizon.net] has joined #sbcl 18:56:47 ASau` [~user@p5797EFC0.dip0.t-ipconnect.de] has joined #sbcl 18:57:15 -!- ASau [~user@p5797EFC0.dip0.t-ipconnect.de] has quit [Remote host closed the connection] 19:02:34 -!- slyrus [~chatzilla@adsl-99-183-240-185.dsl.pltn13.sbcglobal.net] has quit [Ping timeout: 252 seconds] 19:03:54 -!- ASau` [~user@p5797EFC0.dip0.t-ipconnect.de] has quit [Remote host closed the connection] 19:04:28 ASau` [~user@p5797EFC0.dip0.t-ipconnect.de] has joined #sbcl 19:07:59 yacks [~py@180.151.36.168] has joined #sbcl 19:55:34 slyrus [~chatzilla@adsl-99-183-240-185.dsl.pltn13.sbcglobal.net] has joined #sbcl 19:57:38 there's something really really wrong about lvar-derive-type. 19:57:52 It's only shown by the new VOPs, but they should never be selected in the first place. 19:59:27 Bisecting a bit to see where the actual breakage happened. Hopefully it's on me. 20:01:02 CL-USER> (sb-c::single-value-type 20:01:02 (sb-c::values-type-union (sb-c::values-specifier-type '(values character &optional)) 20:01:05 (sb-c::values-specifier-type '(values &optional)))) 20:01:08 # 20:01:11 wut? 20:07:28 -!- slyrus [~chatzilla@adsl-99-183-240-185.dsl.pltn13.sbcglobal.net] has quit [Ping timeout: 252 seconds] 20:09:38 scymtym_ [~user@ip-5-147-122-209.unitymediagroup.de] has joined #sbcl 20:15:44 i tried to use cl-test-grid (of which i am a novice user) to find regression from 1.1.7 to current master on x86 and it produced the following result: http://www.techfak.uni-bielefeld.de/~jmoringe/sbcl-test-grid/demo/sbcl-diff.html (SBCL versions are stuck at 1.0.57 in my tree, but the commit hash are correct) 20:18:25 cl-csv: (funcall '(setf ...) ...) is broken code 20:19:35 pkhuong: there may very well be false positives 20:19:47 i made no attempt to verify the results 20:20:19 actually, the runs just finished and i thought more eyes would be better than fewer 20:21:33 I'm jsut trying to go through them 20:21:49 cl-geo was elided, so I have no clue what went wrong :\ 20:22:22 cl-locative is cute ;) 20:27:26 yeah. So I'll fix single-value-type to type-union NULL in. That should fix the MOVE VOP issues. There's a bunch of failures cause by (funcall '(setf ...) ...) in clsql. A lot of 404. And the rest, I don't have neough logs :\ 20:27:40 trivial-timeout-test might be a regression. 20:30:29 pkhuong: i realized the connectivity-related false positives just now, sorry 20:32:48 same for hu.dwim.reiterate.test 20:40:12 ASau`` [~user@p5797EA99.dip0.t-ipconnect.de] has joined #sbcl 20:41:11 -!- ASau` [~user@p5797EFC0.dip0.t-ipconnect.de] has quit [Ping timeout: 240 seconds] 20:42:13 davazp [~user@178.167.175.194.threembb.ie] has joined #sbcl 20:56:17 pkhuong: manually logs for geo: http://www.techfak.uni-bielefeld.de/~jmoringe/sbcl-test-grid/geo.txt 20:56:28 manually generated* 20:57:27 -!- ASau`` is now known as ASau 21:01:07 hm. I wouldn't have said it was likely that handling of floating point had changed dramatically 21:02:25 in 0b3f5cc, it seems to be sufficient to try (compile nil (lambda () (asin -1.0f0))) 21:03:29 x86? did libm change in the meantime? 21:03:34 (with safety 3 and debug 3) 21:04:42 pkhuong: x86: yes, libm: between when and when? 21:05:02 failure and success 21:05:37 no change that i am aware of 21:05:47 i will try to investigate 21:05:58 Krystof: do you agree that single-value-type is borked, or am I missing something? 21:06:35 it certainly seems borked 21:06:47 another one for the "how has this worked for so long?" 21:07:00 http://paste.lisp.org/display/137343 21:08:01 *scymtym_* can't find a libm update between the test runs 21:08:38 scymtym_: do you think you'll have time to bisect in the next couple days? 21:09:00 pkhuong: sure 21:09:05 thanks 21:09:07 i will need guidance though 21:09:57 this conference has really bad wifi, as usual. I'm not even sure I can commit the s-v-type fix. 21:11:59 do you think turning the cl-test-grid stuff into a maintainer script would be useful? (i think, ecl already has some kind of automation for this) 21:12:06 yes! 21:12:29 Krystof: ok, i will try then 21:12:49 it would be awesome 21:12:51 will most likely not be ready before this release, though 21:13:16 even better than the autobench stuff 21:13:26 and it takes ~ 6 - 8 hours to run :) 21:13:30 pkhuong: ok, values-type-union is imprecise, so any time we get a values type and ask for its single value we must in principle be ready for NIL in addition to whatever we get 21:13:38 scymtym_: running it once a week would be plenty good 21:14:45 (does autobench still run anywhere?) 21:14:46 i got my bosses permission to run SBCL jobs on our test servers today 21:15:04 Krystof: well, I think it's not imprecise as much as correct for a values type 21:15:04 maybe the test-grid stuff can run there 21:16:21 next step: automated bisection on test-grid failures? (: 21:17:11 interesting idea 21:17:49 6-8 hours per run is too slow to run it per commit, but if it runs daily or so, bisection should be computationally doable. 21:18:12 -!- prxq [~mommer@mnhm-590c19f0.pool.mediaWays.net] has quit [Remote host closed the connection] 21:18:40 the bisection would not have to run on "quicklisp libworld", the failing system would be sufficient 21:18:45 right 21:19:02 (and on new successes, actually. unexpected successes are sometimes more worrying than failures ;) 21:19:39 the amount of networking involved might get in the way of that much automation though 21:20:49 in theory a caching proxy would work, but I'm in no position to suggest projects on that scale 21:22:21 cl-test-grid by default has cloud storage of all results required uploads for test runs and downloads for evaluation (if i understand it correctly) 21:22:29 not sure how easily that can be cached 21:22:50 but maybe local storage is possible as well 21:23:30 -!- leoc [~leoc.git@p5480A412.dip0.t-ipconnect.de] has quit [Remote host closed the connection] 21:24:14 -!- scymtym_ [~user@ip-5-147-122-209.unitymediagroup.de] has quit [Remote host closed the connection] 21:25:06 scymtym_ [~user@ip-5-147-122-209.unitymediagroup.de] has joined #sbcl 21:27:24 if someone's up for it, values type union could be optimised by computing the number of types to unite before taking their union... rather than throwing stuff away after the fact. 21:42:42 Given the amount of activity so far this freeze period, are we still considering releasing before the first? 21:44:36 true. a downloadable rc sounds like a better idea. 22:28:25 stassats [~stassats@wikipedia/stassats] has joined #sbcl 22:31:53 -!- foreignFunction [~niksaak@ip-4761.sunline.net.ua] has quit [Quit: Leaving.] 22:38:28 VOP selection is independent of each other? 22:39:36 so, once the template is decided that it's ok, it can't be rescinded in case if the other template could benefit from a different version of the first template 22:45:36 yeah. 22:49:14 It should be doable to detect when move vops + vop is cheaper than a vop with fewer/no move. 22:49:38 *pkhuong* bets there's a dozen such nonsense costs. 22:50:03 i have no faith in :costs for detecting what's cheaper 22:51:21 http://www.techfak.uni-bielefeld.de/~jmoringe/run-test-grid.sh 22:51:26 can be placed in root of SBCL tree and run from there; reporting at the end has to be tweaked 22:51:39 so, maybe some vop families should be marked as complete, meaning whichever you choose, it's the best, then just relying on reducing the number of converting moves 22:51:41 stassats: pretty sure something like tree tiling at the IR2 level would be easier to reason about. 22:52:11 so we can mark a template as equivalent (better) than a given tree of templates 22:52:34 i was looking for some low hanging fruits 22:53:23 and getting to know ltn, who would've thought that "ltn" could be the name for the thing which does instruction selection 22:53:38 there's that (: 22:54:46 *stassats* is glad there's some learning progress, some things are now more familiar 22:55:26 now, how would i write documentation for people aren't familiar with it, if it all looks "obvious" to me, i should have started from the beginning and in parallel 22:56:14 i wouldn't be able to ask stupid questions when i'm able to answer them 22:56:57 maybe that's why the compiler is so poorly documented 23:04:05 there's the option of writing complete answers instead of/as well as answering one-off questions on irc or by email. 23:06:28 i often have a problem with abstract parts, i guess the best documentation would be in a form which have source code examples, and describing how that code affects that particular stage of the compiler 23:07:21 like "what does stack analysis do? oh, it's when (multiple-value-bind (a b c) .... (return-from ...)) happens" 23:07:50 not "it's used to insert clean up forms for the stack when NLX happens" 23:10:47 and a set of answers to questions like "why is this not done some other way?" 23:10:58 "I wish I knew" ;) 23:12:55 and sbcl-internals.cliki is still down, i guess we need some other place for a wiki 23:14:31 and i shouldn't poke sbcl at night, it causes bad dreams 23:20:34 -!- stassats [~stassats@wikipedia/stassats] has quit [Ping timeout: 276 seconds] 23:22:20 trivial timeout looks like a version mismatch with lift. 23:29:18 so, re bound-func: it's broken (fl(f(x-eps)) might be eql fl(f(x)), even if f is strictly monotonic), but I still don't see what the new problem is. 23:48:12 -!- Quadrescence [~quad@unaffiliated/quadrescence] has quit [Quit: Leaving]