2016-09-01T00:01:27Z cromachina joined #sbcl
2016-09-01T00:01:42Z DeadTrickster quit (Ping timeout: 244 seconds)
2016-09-01T00:05:08Z DeadTrickster joined #sbcl
2016-09-01T00:14:07Z DeadTrickster quit (Ping timeout: 244 seconds)
2016-09-01T00:14:42Z oleo_ joined #sbcl
2016-09-01T00:18:30Z oleo quit (Ping timeout: 265 seconds)
2016-09-01T00:31:32Z sjl quit (Read error: Connection reset by peer)
2016-09-01T00:47:43Z stassats: scymtym: all but one internal error actually have fixed arguments
2016-09-01T01:54:01Z em1l joined #sbcl
2016-09-01T01:57:27Z em1l_ quit (Ping timeout: 244 seconds)
2016-09-01T02:48:07Z rpg left #sbcl
2016-09-01T02:52:49Z stassats: removing arg number from errors, saves 65K
2016-09-01T03:00:51Z stassats: can save one more byte, the trap number is 0-8, can just assume any trap byte greater than 8 is of kind error
2016-09-01T03:06:47Z stassats: removing arg length from x86-64 sorta works, everything else needs to be adjusted
2016-09-01T03:06:56Z stassats: will handle later
2016-09-01T03:12:30Z nyef: stassats: Did you just rebase my read-only-tramps branch and build from that, or did it require some tweaking?
2016-09-01T03:14:51Z stassats quit (Ping timeout: 264 seconds)
2016-09-01T04:15:23Z oleo_ quit (Changing host)
2016-09-01T04:15:24Z oleo_ joined #sbcl
2016-09-01T04:15:26Z oleo_ is now known as oleo
2016-09-01T04:50:27Z oleo quit (Quit: Leaving)
2016-09-01T05:03:08Z pootler_ quit (Ping timeout: 250 seconds)
2016-09-01T05:04:47Z pootler_ joined #sbcl
2016-09-01T05:13:30Z shka_ joined #sbcl
2016-09-01T05:34:53Z angavrilov joined #sbcl
2016-09-01T05:58:08Z myrkraverk: I am trying to build 1.3.9 on OS X and get this:
2016-09-01T05:58:10Z myrkraverk: ld: library not found for -lgcc_s.10.4
2016-09-01T05:58:33Z myrkraverk: Is that a regression from 1.3.8?  (which I installed with mac ports)
2016-09-01T06:04:55Z shka_ quit (Ping timeout: 244 seconds)
2016-09-01T06:10:49Z myrkraverk: Ok, I managed to work around it.
2016-09-01T06:23:25Z myrkraverk quit (Ping timeout: 265 seconds)
2016-09-01T06:24:53Z myrkraverk joined #sbcl
2016-09-01T06:26:33Z scymtym quit (Ping timeout: 276 seconds)
2016-09-01T06:30:57Z myrkraverk quit (Remote host closed the connection)
2016-09-01T07:03:00Z myrkraverk joined #sbcl
2016-09-01T07:21:00Z myrkraverk: Ok, so I'm back, and still trying to build SBCL 1.3.9 on OS X.  I worked around the previous error with
2016-09-01T07:21:18Z myrkraverk: SBCL_MACOSX_VERSION_MIN=10.10 sh make.sh
2016-09-01T07:21:29Z myrkraverk: And that seems to work, except now it always ends with
2016-09-01T07:21:54Z myrkraverk: make-target-2.sh: line 48: 29934 Killed: 9               ./src/runtime/sbcl --core output/cold-sbcl.core --lose-on-corruption --no-sysinit --no-userinit < make-target-2.lisp
2016-09-01T07:22:07Z myrkraverk: Is there some sort of timeout in the build system that I can turn off?
2016-09-01T07:22:16Z myrkraverk: It does not appear to be a memory kill.
2016-09-01T07:24:04Z scymtym joined #sbcl
2016-09-01T07:31:11Z myrkraverk: Is there any way to fix my build error/problem?
2016-09-01T07:34:35Z Xof: I don't think it's a timeout
2016-09-01T07:34:47Z Xof: what signal is number 9 on OS X?
2016-09-01T07:34:59Z Xof: did building sbcl-1.3.8 work?
2016-09-01T07:35:06Z Xof: if so, then it might be bisection time :-/
2016-09-01T07:35:08Z brucem: Xof: 9 is kill.
2016-09-01T07:51:18Z attila_lendvai joined #sbcl
2016-09-01T07:51:19Z attila_lendvai quit (Changing host)
2016-09-01T07:51:19Z attila_lendvai joined #sbcl
2016-09-01T07:56:09Z scymtym: minion: memo for stassats: latest version of my changes is https://github.com/scymtym/sbcl/commits/wip-compact-error-trap-args . which parts still make sense given that you have argument counts already covered?
2016-09-01T07:56:10Z minion: Remembered. I'll tell stassats when he/she/it next speaks.
2016-09-01T07:57:01Z myrkraverk: I have 1.3.8 from mac ports.
2016-09-01T07:57:11Z myrkraverk: But yeah, I'll try to build that too, just to see if that works.
2016-09-01T08:10:56Z Xof: can someone remind me whether RPi is armhf or softfp?
2016-09-01T08:10:59Z myrkraverk: I'm attempting 1.3.8 in exactly the same way as before.
2016-09-01T08:12:58Z jackdaniel: Xof: softfp
2016-09-01T08:14:31Z jackdaniel: oh, no, it's armhf, sorry
2016-09-01T08:15:04Z Xof:   Flags:                             0x5000402, has entry point, Version5 EABI, hard-float ABI
2016-09-01T08:15:15Z Xof: don't worry; I think the answer was "it depends"
2016-09-01T08:15:18Z Xof: but I got there in the end :-)
2016-09-01T08:15:21Z Xof: thanks
2016-09-01T08:26:59Z eschatologist quit (Ping timeout: 260 seconds)
2016-09-01T08:27:10Z eschatologist joined #sbcl
2016-09-01T08:48:17Z DavidGu joined #sbcl
2016-09-01T08:58:49Z Bike quit (*.net *.split)
2016-09-01T08:58:49Z karswell quit (*.net *.split)
2016-09-01T08:58:50Z Shinmera quit (*.net *.split)
2016-09-01T08:58:50Z irsol quit (*.net *.split)
2016-09-01T08:58:51Z mood quit (*.net *.split)
2016-09-01T09:03:04Z Bike joined #sbcl
2016-09-01T09:03:04Z karswell joined #sbcl
2016-09-01T09:03:04Z Shinmera joined #sbcl
2016-09-01T09:03:04Z irsol joined #sbcl
2016-09-01T09:03:04Z mood joined #sbcl
2016-09-01T09:03:07Z Bike quit (Quit: sleep)
2016-09-01T09:27:53Z ASau quit (Ping timeout: 258 seconds)
2016-09-01T09:48:59Z DeadTrickster joined #sbcl
2016-09-01T10:18:43Z stassats joined #sbcl
2016-09-01T10:24:41Z stassats: scymtym: i think the top commit will be subsumed by elimination of arg-length
2016-09-01T10:24:41Z minion: stassats, memo from scymtym: latest version of my changes is https://github.com/scymtym/sbcl/commits/wip-compact-error-trap-args . which parts still make sense given that you have argument counts already covered?
2016-09-01T10:26:01Z stassats: scymtym: can you make a version of read-var-integer that works on saps?
2016-09-01T10:44:49Z Xof: stassats: did you build a PPC binary?
2016-09-01T10:45:23Z stassats: it's a shared machine, so i'm not too keen using their gcc
2016-09-01T10:46:48Z stassats: i only build windows binaries because it's hard to build on windows (and they are the most popular ones)
2016-09-01T10:47:16Z stassats: and arm64
2016-09-01T10:50:00Z stassats: https://sourceforge.net/projects/sbcl/files/sbcl/1.2.7/sbcl-1.2.7-powerpc-linux-binary.tar.bz2/stats/timeline?dates=2015-01-01+to+2016-09-01
2016-09-01T10:51:35Z Xof: heh
2016-09-01T11:18:50Z sjl joined #sbcl
2016-09-01T11:18:56Z scymtym: stassats: sure, will have to wait until evening, though
2016-09-01T11:19:26Z stassats: i have a hack in place to proceed without it, so, no problem
2016-09-01T11:20:01Z stassats: besides, it's going to take me a day to fully transition to arg-count-less traps
2016-09-01T11:20:20Z stassats: and then we need full read-only-tramps support before your changes can be merged anyway
2016-09-01T11:21:48Z scymtym: i have to do lots of cleaning up and commenting as well
2016-09-01T11:27:40Z myrkraverk: make-target-2.sh: line 48: 31036 Killed: 9               ./src/runtime/sbcl --core output/cold-sbcl.core --lose-on-corruption --no-sysinit --no-userinit < make-target-2.lisp
2016-09-01T11:27:51Z myrkraverk: happens also with 1.3.8 -- so maybe it is a memory problem.
2016-09-01T11:28:05Z myrkraverk: but that can only happen if sbcl is compiling a 32 bit process.
2016-09-01T11:28:26Z stassats: compiling a process?
2016-09-01T11:28:44Z schjetne joined #sbcl
2016-09-01T11:30:27Z myrkraverk: Compiling SBCL itself.
2016-09-01T11:31:18Z myrkraverk: I had the exact same error with 1.3.9 before.
2016-09-01T11:31:19Z attila_lendvai quit (Ping timeout: 260 seconds)
2016-09-01T11:31:38Z myrkraverk: So I thought to check (or it was a suggestion) if it was a regression.
2016-09-01T11:31:42Z myrkraverk: I'm on OS X, btw.
2016-09-01T11:33:16Z stassats: (list (lisp-implementation-version) (machine-type) (software-type) (software-version)) => ("1.3.8.47-f8b76ac" "X86-64" "Darwin" "15.6.0")
2016-09-01T11:33:51Z myrkraverk: Mine is ("1.3.8" "X86-64" "Darwin" "14.5.0")
2016-09-01T11:34:13Z myrkraverk: From macports as far as I can remember.
2016-09-01T11:35:01Z myrkraverk: But I'm trying to update to 1.3.9 from the source code.
2016-09-01T11:35:20Z myrkraverk: Incidentally, a coworker had no trouble with that.
2016-09-01T11:36:52Z myrkraverk: % file src/runtime/sbcl
2016-09-01T11:36:53Z myrkraverk: src/runtime/sbcl: Mach-O 64-bit executable x86_64
2016-09-01T11:37:02Z myrkraverk: It's 64bit alright.
2016-09-01T11:37:24Z stassats: incidentally, sb-qshow is broken on x86-darwin
2016-09-01T11:37:44Z myrkraverk: And is sb-qshow what I'd use to "debug" this problem?
2016-09-01T11:38:24Z stassats: well, it could help, but i have it always enabled and couldn't build x86-darwin
2016-09-01T11:39:21Z myrkraverk: I see.
2016-09-01T11:39:28Z myrkraverk: How do I enable it?
2016-09-01T11:39:39Z stassats: --with-qshow
2016-09-01T11:39:51Z stassats: or --with-sb-qshow, rather
2016-09-01T11:40:32Z myrkraverk: Ok.
2016-09-01T11:41:25Z stassats: arrange_return_to_c_function is atrocious
2016-09-01T11:42:22Z p_l quit (Remote host closed the connection)
2016-09-01T11:44:01Z stassats: well, the current x86 git is broken completely
2016-09-01T11:44:21Z stassats: 1.3.9 can get built
2016-09-01T11:44:40Z myrkraverk:  sbcl	17.50 GB	9.44 GB	2	11	32364	johann	96.2	1:10.15	0	0 bytes	0 bytes	64 bit		-	No	No	0 bytes	0 bytes	0	0	0 bytes	0 bytes	0 bytes	No	No		0 bytes	
2016-09-01T11:44:47Z myrkraverk: Sorry about the overload.
2016-09-01T11:44:54Z myrkraverk: So, maybe it /is/ a memory kill.
2016-09-01T11:45:04Z stassats: bloody hell, i can't cross build x86-linux anymore
2016-09-01T11:45:05Z stassats: stupid debian
2016-09-01T11:45:06Z myrkraverk: I only have 16GB on this machie.
2016-09-01T11:45:13Z stassats: "only"
2016-09-01T11:45:32Z myrkraverk: In this case, it really feels like "only"
2016-09-01T11:46:01Z myrkraverk: The process is using almost 40GB now.
2016-09-01T11:46:29Z myrkraverk: I'm going to check if macports have a patch.
2016-09-01T11:46:40Z stassats: let's recap, that is building which sbcl version? with which sbcl version?
2016-09-01T11:46:43Z stassats: for what arch
2016-09-01T11:47:23Z myrkraverk: Host SBCL: 1.3.8, OS X.  Target SBCL: 1.3.9, OS X.
2016-09-01T11:47:34Z stassats: both x86-64?
2016-09-01T11:47:47Z stassats: well, must be a kernel bug then
2016-09-01T11:48:14Z myrkraverk: Yes.
2016-09-01T11:48:22Z myrkraverk: Could very well be.
2016-09-01T11:48:37Z myrkraverk: I'll try with qshow and see if anything changes.
2016-09-01T11:48:48Z stassats: well, qshow has to be enabled and stuff
2016-09-01T11:48:52Z stassats: at runtime
2016-09-01T11:49:18Z myrkraverk: Oh.
2016-09-01T11:49:24Z stassats: so, can you do ./src/runtime/sbcl --core output/cold-sbcl.core --lose-on-corruption --no-sysinit --no-userinit --load make-target-2.lisp
2016-09-01T11:49:32Z stassats: and when it starts to behave strangely, interrupt it
2016-09-01T11:49:37Z stassats: and get a backtrace
2016-09-01T11:50:00Z myrkraverk: I'll try that in a bit.
2016-09-01T11:51:12Z stassats: i'll try fixing the x86 build (HEAD, not yours)
2016-09-01T11:51:23Z myrkraverk: Haha, fair enough.
2016-09-01T11:56:02Z stassats: ok, that's done
2016-09-01T11:56:19Z p_l joined #sbcl
2016-09-01T11:56:27Z stassats: now for restoring the sb-qshow build on darwin-x86
2016-09-01T11:59:34Z myrkraverk: I can't seem to be able to interrupt sbcl now.
2016-09-01T11:59:42Z myrkraverk: I just get several ^Cs
2016-09-01T12:00:04Z rpg joined #sbcl
2016-09-01T12:01:18Z stassats: try with kill
2016-09-01T12:02:09Z myrkraverk: "nothing happens" with kill <pid>
2016-09-01T12:03:56Z myrkraverk: The return value from the killed SBCL is 137.
2016-09-01T12:04:18Z myrkraverk: I'm mentioning it on the off chance it tells something.
2016-09-01T12:04:37Z stassats: it means it was killed
2016-09-01T12:04:49Z myrkraverk: *nod*
2016-09-01T12:04:55Z stassats: (- 137 128) => 9
2016-09-01T12:05:04Z myrkraverk: Ah.
2016-09-01T12:05:16Z DGASAU quit (Read error: Connection reset by peer)
2016-09-01T12:05:58Z myrkraverk: So I it uses all the memory until killed, and I can't interrupt it to find out why.  Yay.
2016-09-01T12:06:59Z DGASAU joined #sbcl
2016-09-01T12:10:16Z myrkraverk: Will a debugger generated backtrace help to figure out what the problem might be?
2016-09-01T12:10:19Z myrkraverk: because I have one now.
2016-09-01T12:10:34Z stassats: wouldn't hurt
2016-09-01T12:10:55Z stassats: you can also call ldb_monitor() from the debugger
2016-09-01T12:11:06Z stassats: and poke around there, 'backtrace' and whatnot
2016-09-01T12:11:54Z myrkraverk: http://paste.lisp.org/+6YLT
2016-09-01T12:13:25Z stassats: if you reattach again, is it still there?
2016-09-01T12:14:13Z stassats: i can send you a 1.3.9 binary
2016-09-01T12:14:32Z myrkraverk: I never detached, and running ldb_monitor() crashed it.
2016-09-01T12:14:48Z myrkraverk: Yeah, I'll take a 1.3.9 binary.  How do you want to send it?
2016-09-01T12:14:58Z stassats: pigeons
2016-09-01T12:15:52Z myrkraverk: Hmm, there's a risk that carrier pigeons will be killed and eaten on the way.
2016-09-01T12:15:58Z myrkraverk: I live in a high risk area for birds.
2016-09-01T12:16:27Z rpg quit (Quit: My Mac has gone to sleep. ZZZzzz…)
2016-09-01T12:16:38Z myrkraverk: Btw, I updated the past with the failure output at running ldb_monitor()
2016-09-01T12:16:42Z myrkraverk: *paste
2016-09-01T12:16:55Z DGASAU quit (Ping timeout: 244 seconds)
2016-09-01T12:17:08Z myrkraverk: Maybe not a good idea to run that while stopped inside libsystem_platform.dylib`_platform_bzero$VARIANT$Ivybridge + 120
2016-09-01T12:17:22Z myrkraverk: deep inside calloc.
2016-09-01T12:18:14Z stassats: are you building threaded?
2016-09-01T12:18:14Z myrkraverk: But it did print this: Welcome to LDB, a low-level debugger for the Lisp runtime environment.
2016-09-01T12:18:26Z myrkraverk: I don't know.
2016-09-01T12:19:35Z myrkraverk: Is the stage 1 sbcl ever threaded?
2016-09-01T12:19:44Z myrkraverk: Because I can run that and find out, I guess.
2016-09-01T12:20:18Z myrkraverk: Or maybe not, since I don't have a core file yet.
2016-09-01T12:22:23Z stassats: myrkraverk: https://drive.google.com/open?id=0B0Xjfi08W3JmaGwtTDB4RGdTT1k
2016-09-01T12:23:54Z DGASAU joined #sbcl
2016-09-01T12:25:18Z myrkraverk: stassats: thank you.  Can I install that in the usual way?
2016-09-01T12:26:58Z myrkraverk: Well, run-sbcl.sh is working, so I assume so.  Thank you very much.
2016-09-01T12:34:03Z schjetne quit (Ping timeout: 265 seconds)
2016-09-01T12:35:33Z DGASAU quit (Read error: Connection reset by peer)
2016-09-01T12:35:59Z DGASAU joined #sbcl
2016-09-01T12:54:15Z DGASAU quit (Read error: Connection reset by peer)
2016-09-01T12:54:41Z DGASAU joined #sbcl
2016-09-01T12:58:39Z DougNYC joined #sbcl
2016-09-01T13:27:43Z jzp quit (Ping timeout: 244 seconds)
2016-09-01T13:27:51Z jzp joined #sbcl
2016-09-01T13:48:42Z Posterdati: hi
2016-09-01T13:49:00Z Posterdati: how can I install sbcl and its sources?
2016-09-01T13:49:20Z Posterdati: I'm using the build-sbcl.sh script
2016-09-01T13:53:29Z stassats: what is build-sbcl.sh?
2016-09-01T13:53:54Z stassats: anyhow, just read the documentation
2016-09-01T14:04:19Z oleo joined #sbcl
2016-09-01T14:05:47Z cromachina quit (Read error: Connection reset by peer)
2016-09-01T14:20:52Z CORDIC quit (Ping timeout: 265 seconds)
2016-09-01T14:21:49Z rpg joined #sbcl
2016-09-01T14:26:07Z myrkraverk: Question, is there anything inside SBCL that I can use to create a debugger plugin for lldb?
2016-09-01T14:26:39Z myrkraverk: I mean, like (formerly Sun's) dbx could (can?) step between C and Java code in mixed applications.
2016-09-01T14:27:29Z myrkraverk: I just thought ti could be a fun project, if there's already something in SBCL I can hook into -- is LDB what I want to hook into?
2016-09-01T14:27:46Z pkhuong: I'd start by getting backtraces working
2016-09-01T14:28:26Z pkhuong: actually, scratch that. I'd start by making code not use a copying GC
2016-09-01T14:28:37Z myrkraverk: You mean, that when I do "thread backtrace" or the equivalent in lldb, I get lisp backtraces?
2016-09-01T14:28:42Z rpg quit (Quit: My Mac has gone to sleep. ZZZzzz…)
2016-09-01T14:29:21Z myrkraverk: What do you mean by a "not copying GC" -- do I need to fix SBCL GC before I can do this?
2016-09-01T14:29:38Z pkhuong: yes. or mixed C -> Lisp if you're in the middle of a foreign function call
2016-09-01T14:29:48Z pkhuong: you would need some GC work first, indeed.
2016-09-01T14:30:08Z myrkraverk: But mixed C and Lisp already works with cffi.
2016-09-01T14:30:33Z myrkraverk: So, I'm at a bit of a loss about what the issue exactly is -- note that I don't know SBCL's internals yet.
2016-09-01T14:30:42Z pkhuong: was not aware CFFI did backtraces in lldb
2016-09-01T14:30:50Z myrkraverk: Oh, I'm sure it doesn't.
2016-09-01T14:31:05Z myrkraverk: So I guess I'm completely misunderstand the issue.
2016-09-01T14:31:34Z pkhuong: if you want to integrate SBCL and LLDB better, I think getting backtraces to work will be a good stepping stone.
2016-09-01T14:31:46Z pkhuong: however, the unix toolchain tends to assume code doesn't move around.
2016-09-01T14:32:27Z myrkraverk: Ah.  I don't know about lldb -- but if I run into issues with it, I can always "fix" lldb too.
2016-09-01T14:32:30Z pkhuong: you can register new sections of code (for JITting purposes), and sometimes remove them, but the none of it is optimised to handle moves.
2016-09-01T14:33:07Z myrkraverk: Ah, I see.  So lldb is probably not prepared to meet code that moves around.
2016-09-01T14:33:11Z pkhuong: seems simpler to fix the way SBCL allocates code than to dive deep in LLDB and try to get them to accept patches for a fringe language.
2016-09-01T14:33:25Z myrkraverk: Right.
2016-09-01T14:34:53Z myrkraverk: I guess I'll try to add some research into this project into my copious free time.
2016-09-01T14:47:58Z scymtym quit (Ping timeout: 258 seconds)
2016-09-01T14:54:22Z DGASAU quit (Read error: Connection reset by peer)
2016-09-01T14:55:01Z DGASAU joined #sbcl
2016-09-01T14:56:21Z dougk_: pkhuong: and myrkraverk: i'm about to commit a code-never-moves patch after a little bit more performance testing here. It's a separately managed bag of pages that use a best-fit allocator that plays nicely with gencgc
2016-09-01T15:02:10Z myrkraverk: Nice, then I don't have to recreate that work.
2016-09-01T15:02:36Z myrkraverk: And can delve directly into stacktraces in lldb.
2016-09-01T15:03:09Z myrkraverk: I know that in dbx, it was (is?) done with a plugin that runs inside the JVM.
2016-09-01T15:03:49Z myrkraverk: My hope was that SBCL has hooks so I don't need a plugin -- and of course I can just edit SBCL itself too.
2016-09-01T15:05:10Z myrkraverk: dougk_: do you have an approximate date for that commit?
2016-09-01T15:09:12Z sjl quit (Ping timeout: 265 seconds)
2016-09-01T15:14:31Z stassats: code on the stack doesn't move anyway
2016-09-01T15:14:38Z stassats: on x86
2016-09-01T15:14:55Z gargaml joined #sbcl
2016-09-01T15:16:20Z rpg joined #sbcl
2016-09-01T15:18:46Z sjl joined #sbcl
2016-09-01T15:32:29Z rpg_ joined #sbcl
2016-09-01T15:33:12Z rpg quit (Ping timeout: 276 seconds)
2016-09-01T15:34:15Z rpg_ is now known as rpg
2016-09-01T15:37:10Z dougk_: myrkraverk: "Real Soon Now" (tm). Seriously, <= 2 weeks
2016-09-01T15:39:57Z myrkraverk: Ok, that's short enough that I won't have to wait for you.
2016-09-01T15:43:37Z stassats: myrkraverk: so, does my binary run?
2016-09-01T15:44:01Z myrkraverk: Yes, it works flawlessly in my limited tests earlier this evening.
2016-09-01T15:47:18Z stassats: maybe single threaded sbcl fails
2016-09-01T15:50:24Z gingerale joined #sbcl
2016-09-01T15:57:24Z stassats: no, works fine
2016-09-01T15:57:45Z stassats: myrkraverk: which c compiler are you using?
2016-09-01T15:58:03Z myrkraverk: clang
2016-09-01T15:58:20Z stassats: can you try gcc?
2016-09-01T15:58:32Z myrkraverk: Not tonight, I'm about to sleep.
2016-09-01T15:59:28Z rpg quit (Quit: My Mac has gone to sleep. ZZZzzz…)
2016-09-01T16:01:35Z stassats: now i need to test cold init errors
2016-09-01T16:01:43Z stassats: the dreaded cold init errors
2016-09-01T16:02:34Z rpg joined #sbcl
2016-09-01T16:06:33Z rpg quit (Ping timeout: 240 seconds)
2016-09-01T16:10:55Z scymtym joined #sbcl
2016-09-01T16:17:47Z stassats: close to getting arg-count-less errors to work
2016-09-01T16:17:49Z stassats: everywhere
2016-09-01T16:30:21Z shka_ joined #sbcl
2016-09-01T16:31:04Z rpg joined #sbcl
2016-09-01T16:43:17Z chris_l joined #sbcl
2016-09-01T16:48:37Z Bike joined #sbcl
2016-09-01T16:58:50Z sjl quit (Read error: Connection reset by peer)
2016-09-01T17:06:52Z stassats: a final round of testing and pushing
2016-09-01T17:07:12Z stassats: then pushing what read-only-tramps are already working and enabling them
2016-09-01T17:16:13Z nyef: Sounds good. At that point I can start in on the MIPS version.
2016-09-01T17:16:44Z nyef: ... And a branch for getting rid of mips-assem.S, as well.
2016-09-01T17:16:47Z stassats: well, i can handle mips, maybe you can do sparc/alpha/hppa?
2016-09-01T17:17:14Z nyef: Okay, yeah, I can handle sparc/alpha/hppa... If I can remember how to get the alpha working again.
2016-09-01T17:19:25Z stassats: 41 files changed, 189 insertions(+), 309 deletions(-)
2016-09-01T17:20:46Z stassats: i think for today i'll bring ppc and x86 in line with the 64oids and push
2016-09-01T17:21:08Z nyef: Sounds good to me.
2016-09-01T17:21:33Z stassats: testing on mips is slow
2016-09-01T17:23:59Z stassats: ok, i'll push before mips finishes building and will handle any fallout afterwards
2016-09-01T17:24:38Z stassats: accidentally interrupted tests on ppc, damnit
2016-09-01T17:26:27Z dustinm` quit (Quit: Leaving)
2016-09-01T17:27:54Z gargaml quit (Quit: WeeChat 1.5)
2016-09-01T18:01:35Z dustinm` joined #sbcl
2016-09-01T18:09:36Z eschatologist: Has SBCL considered moving to non-SourceForge hosting? Grabbing a binary distribution with curl tends not to follow the redirects SourceForge puts in.
2016-09-01T18:10:38Z stassats: don't grab it with curl
2016-09-01T18:10:43Z stassats: then
2016-09-01T18:10:56Z eschatologist: Not all environments have wget out of the box.
2016-09-01T18:11:13Z schjetne joined #sbcl
2016-09-01T18:11:27Z stassats: tough luck
2016-09-01T18:12:54Z joshe: curl -LO works fine for me
2016-09-01T18:14:06Z eschatologist: joshe: Cool, thank you.
2016-09-01T18:14:33Z joshe: curl doesn't follow redirects without -L
2016-09-01T18:15:12Z joshe: probably some sort of backwards compat thing like defaulting to downloading to stdout
2016-09-01T18:15:22Z eschatologist: Giving it a try now, it looks to be working. Thanks!
2016-09-01T18:35:12Z schjetne quit (Ping timeout: 276 seconds)
2016-09-01T18:36:32Z stassats: huh, weird read-only-tramps error on ppc, it tries to LOAD tramps.lisp
2016-09-01T18:36:57Z stassats: and DEFINE-ASSEMBLY-ROUTINE is undefined
2016-09-01T18:42:34Z schjetne joined #sbcl
2016-09-01T18:43:24Z DougNYC quit
2016-09-01T18:45:14Z stassats: just some misdirection cause be excessive deletion
2016-09-01T18:55:43Z DougNYC joined #sbcl
2016-09-01T18:57:54Z stassats: all those tests and compilations, and i only have four backends on read-only-tramps
2016-09-01T19:09:25Z schjetne quit (Ping timeout: 265 seconds)
2016-09-01T19:26:41Z stassats: ok, read-only-tramps are in
2016-09-01T19:33:35Z nyef: Nice.
2016-09-01T19:33:42Z nyef: I suppose now it's my turn.
2016-09-01T19:44:17Z scymtym: and i was just done rebasing :) but it's ok, i still have some work to do before my changes can go in
2016-09-01T19:50:11Z chris_l quit (Quit: Ex-Chat)
2016-09-01T19:54:21Z DougNYC quit
2016-09-01T20:13:44Z stassats: scymtym: i think now your changes wouldn't cross with future changes
2016-09-01T20:14:21Z stassats: but read-only-tramps needs to be universal before they can go in, unless you want to encode traps by hand
2016-09-01T20:14:36Z stassats: but x86oids, arm64 and ppc are done
2016-09-01T20:15:46Z gingerale quit (Remote host closed the connection)
2016-09-01T20:16:40Z sjl joined #sbcl
2016-09-01T20:18:22Z stassats: i'll delay the trap-number + trap-kind change for after the dust settles
2016-09-01T20:18:47Z stassats: changing code for all backends is not so pleasant
2016-09-01T20:20:44Z Cthulhux: the default windows sbcl doesn't seem to support :compression for save-lisp-and-die, at least i get a PRIMITIVE HALT when trying. annoying. :/ how can i catch that?
2016-09-01T20:21:01Z stassats: catch?
2016-09-01T20:21:30Z Cthulhux: avoid a primitive halt :D
2016-09-01T20:21:47Z stassats: don't pass :compression
2016-09-01T20:21:49Z stassats: easy
2016-09-01T20:22:18Z Cthulhux: why doesn't windows sbcl support compression in the first place and why doesn't it just ignore the flag?
2016-09-01T20:22:55Z stassats: because it requires a shared library
2016-09-01T20:23:02Z stassats: because it actually does ignore it
2016-09-01T20:23:20Z Cthulhux: "ignoring" and "primitive halt" doesn't seem to be identical..
2016-09-01T20:24:05Z stassats: there's no primitive halt for :compression
2016-09-01T20:25:49Z Cthulhux: adding :compression leads to a crash here. hm.
2016-09-01T20:26:36Z mordocai: Cthulhux: does *features* contain :SB-CORE-COMPRESSION on the machine with the crash?
2016-09-01T20:26:45Z mordocai: If it doesn't you might be able to check for that and pass the flag or not on that basis
2016-09-01T20:27:01Z mordocai is not an expert
2016-09-01T20:27:06Z stassats: Cthulhux: you must be doing something wrong then
2016-09-01T20:28:28Z Cthulhux: mordocai, no, it does not. but yes, that could work, thanks...
2016-09-01T20:28:42Z Cthulhux: "doing something wrong" should not crash into LDB, usually. :)
2016-09-01T20:28:58Z mordocai: Cthulhux: putting the full REPL session on pastebin might help as far as determining the cause of the crash, since apparently it shouldn't crash
2016-09-01T20:36:01Z Cthulhux: well, it's reproducible with this one: https://gist.github.com/burtonsamograd/f08f561264ff94391300
2016-09-01T20:36:07Z Cthulhux: add  :compression 9 in line 42 - crash
2016-09-01T20:36:08Z Cthulhux: :/
2016-09-01T20:36:28Z stassats: "tree shaker"
2016-09-01T20:36:32Z stassats: there's your "something wrong"
2016-09-01T20:37:28Z mordocai: Yep. Also, the one comment on that gist talks about your same problem (it says it drops to ldb immediately)
2016-09-01T20:38:07Z mordocai: The comment also states that the image isn't actually smaller
2016-09-01T20:40:42Z stassats: you didn't think to mention first that you cut open the image? and blame :compression instead
2016-09-01T20:40:44Z stassats: sigh
2016-09-01T20:42:08Z stassats: and i had to fire up a windows vm to check that :compression really behaves
2016-09-01T20:42:11Z stassats: double sigh
2016-09-01T20:42:53Z stassats: bug reports, is it that hard to reveal all the pertinent information?
2016-09-01T20:43:07Z angavrilov quit (Remote host closed the connection)
2016-09-01T20:43:48Z mordocai: stassats: Nah, but i've found most people do it badly until they've had considerable experience with it.
2016-09-01T20:44:04Z Cthulhux: without compression it doesn't go into ldb. (and it's indeed smaller for me, by approx. 20 mib.) sorry if this was a total waste of time for you :/
2016-09-01T20:47:40Z mordocai: stassats: That was part my bad too, sorry. I suggested Cthulhux ask here and didn't think to ask for a code snippet first.
2016-09-01T21:12:34Z xristos quit (Quit: ZNC - http://znc.in)
2016-09-01T21:13:13Z shka_ quit (Ping timeout: 258 seconds)
2016-09-01T22:01:15Z rpg quit (Ping timeout: 276 seconds)
2016-09-01T22:08:31Z attila_lendvai joined #sbcl
2016-09-01T22:08:31Z attila_lendvai quit (Changing host)
2016-09-01T22:08:31Z attila_lendvai joined #sbcl
2016-09-01T22:44:19Z rpg joined #sbcl
2016-09-01T23:21:49Z xristos joined #sbcl
2016-09-01T23:46:19Z attila_lendvai quit (Ping timeout: 260 seconds)
2016-09-01T23:58:33Z cromachina joined #sbcl