Camelia, the Perl 6 bug

IRC log for #parrot, 2009-08-30

Parrot | source cross referenced

| Channels | #parrot index | Today | | Search | Google Search | Plain-Text | summary

All times shown according to UTC.

Time Nick Message
00:03 Zak joined #parrot
00:12 GeJ Good morning everyone
00:18 cotto Hi GeJ.
00:38 PacoLinux joined #parrot
00:39 NotFound joined #parrot
01:02 snarkyboojum joined #parrot
01:04 snarkyboojum left #parrot
01:05 snarkyboojum joined #parrot
01:06 NotFound joined #parrot
01:11 rhr joined #parrot
01:53 rhr joined #parrot
02:17 theory joined #parrot
02:24 TiMBuS joined #parrot
02:41 janus joined #parrot
02:58 Zak joined #parrot
03:02 Whiteknight joined #parrot
03:08 Whiteknight good evening #parrot
03:15 jrtayloriv evening
03:18 Whiteknight actually, it's time for bed. Goodnight.
03:20 Coke hurm. 40852 made a segfault go away, even when not using fakeceutable. :|
03:30 rhr joined #parrot
03:45 Coke anything change recently for nans?
03:48 JimmyZ joined #parrot
03:55 dukeleto Coke: what do you mean?
03:57 dalek parrot: r40856 | dukeleto++ | trunk (3 files):
03:57 dalek parrot: [t] Translate most of t/pmc/integer.t into PIR
03:57 dalek parrot: review: https://trac.parrot.org/parrot/changeset/40856/
04:01 Coke had a local mod for tcl, where puts [expr nan] caused a segfault.
04:01 Coke just wondering if any nan guts in parrot changed recently.
04:02 Coke dukeleto: when you added integer-old.t, did you do an svn copy?
04:02 Coke Or just create it new?
04:03 dukeleto coke: i remember your advice, but there is no git svn copy. it is only 2 tests in the -old, though
04:03 Coke *sigh*
04:05 dukeleto Coke: if you would prefer, I can rm the -old and svn copy integer.t to integer-old.t in an svn repo, so that all history is preserved. I just couldn't do it in one go
04:09 dukeleto Coke: it's not that big a deal, as long as the -old files don't linger around very long, correct? Perhaps I should add the necessary things to test_more.pir so that splitting files will not be necessary
04:09 Coke I would have preferred it was done that way originally, because there is no guarante how long those old files are going to be there. I doubt it's worth undoing and doing over.
04:10 * Coke wonders why his vtable override isn't working.
04:13 Coke anyone use gdb find that the source code displayed is all wrong when showing the generated C func?
04:13 Coke (showing chunks of the wrong PIR)
04:13 Coke s/PIR/code/
04:20 dalek parrot: r40857 | dukeleto++ | trunk/t/pmc/integer-old.t:
04:20 dalek parrot: [t][cage] Fix svn properties on t/pmc/integer-old.t
04:20 dalek parrot: review: https://trac.parrot.org/parrot/changeset/40857/
04:29 * dukeleto is reading the source to test_more.pir
04:32 Coke bah. can't convert my PMCs to classes.
04:32 Coke dukeleto: there's not much to it.
04:32 dukeleto Coke: yeah, i meant to say I am reading all it's dependencies and libraries that it pulls in
04:33 dukeleto and t/library/test_more.t
04:44 dukeleto what is the equivalent of eval() in PIR?
04:47 Coke msg cotto PROFILING RUNCORE: Wrote profile to parrot .
04:47 purl Message for cotto stored.
04:48 Coke dukeleto: $P1 = compreg 'PIR'; $P2 = $P1('source code'); $P2()
04:48 dukeleto coke++
04:48 Coke perhaps with some try/catch for good measure.
04:50 dukeleto coke: is there a good example that i can refer to for try/catch? i haven't used it in pir before
04:50 bacek joined #parrot
04:52 dukeleto coke: i am reading docs/user/pir/exceptions.pod, about push_eh and pop_eh, is that what you meant?
04:52 Coke yes.
04:53 Coke see tcl's src/macro : .Try and .TryCatch for some simplistic versions.
04:57 dalek partcl: r648 | coke++ | trunk/tools/profiler.pl:
04:57 dalek partcl: track recent changes to the profiling core in branch.
04:57 dalek partcl: review: http://code.google.com/p/p​artcl/source/detail?r=648
05:03 dukeleto coke: wow, i didn't know you could pass in code to a macro with the { } syntax. i can do all kind of devlish things with that
05:17 cotto Coke, how do you trigger that?
05:19 nopaste "cotto" at 74.61.2.46 pasted "possible fix for Coke" (18 lines) at http://nopaste.snit.ch/17743
05:19 cotto does this help?
05:32 Coke dukeleto: see hllmacros.pir in parrot for more evil.
05:38 dukeleto coke++ # wow
05:43 Coke I avoided it for a long time, but I've been applying it to the tcl code base, I think the resulting code is MUCH more readable.
05:43 Coke of course, it's less debuggable. win some lose some.
05:45 dukeleto good to konw
05:45 cotto Coke, is that patch any better?
05:45 dukeleto know, even
05:49 Coke cotto: haven't tried. moment.
05:52 cotto also, you're using --optimize, aren't you?
05:58 Coke cotto: yes, and yes.
05:58 Coke not sure if the runcore branch is using it, though. checking...
05:59 Coke that branch isn't being config'd with optimize.
06:01 cotto I imagine that it won't help if any compiler optimizations were applied.
06:07 Zak joined #parrot
06:14 cotto yeah.  The hack doesn't work with --optimize
06:16 cotto I guess it'll work to mark it during the runloop, ugly as that is.
06:17 cotto Coke, if you're still up try this:
06:17 nopaste "cotto" at 74.61.2.46 pasted "alternate fix for Coke" (27 lines) at http://nopaste.snit.ch/17744
06:49 mokurai joined #parrot
07:09 sjn joined #parrot
07:15 dukeleto parrot trunk test failures on i386-linux with icu: http://smolder.plusthree.com/app/pu​blic_projects/report_details/26721
07:44 bacek joined #parrot
07:57 iblechbot joined #parrot
08:04 dalek parrot: r40858 | bacek++ | branches/context_pmc3/t/co​mpilers/pge/06-grammar.t:
08:04 dalek parrot: [cage] Initialise result in  t/compilers/pge/06-grammar.t
08:04 dalek parrot: review: https://trac.parrot.org/parrot/changeset/40858/
08:08 bacek joined #parrot
08:09 bacek o hai
08:10 mikehh howdy there
08:11 mikehh All tests PASS (pre/post-config, smoke, nqp_test, fulltest) at r40857 - Ubuntu 9.04 amd64 (g++)
08:11 bacek mikehh: only zombies? :)
08:12 mikehh :-} - whole lot of more karma for g commin' up
08:12 bacek mikehh: O! Can you test context_pmc3 on amd64?
08:13 mikehh 'k - soon as I have finished my current test run
08:13 mikehh let me set it up
08:15 bacek mikehh: thanks!
08:18 mikehh rakudo (7666e92) builds on parrot r40857, make test PASS / make spectest (up to r28122) PASS - Ubuntu 9.04 amd64 (g++)
08:18 dalek parrot: r40859 | bacek++ | branches/context_pmc3/t/pmc/exceptionhandler.t:
08:18 dalek parrot: [cage] Add explicit test for ExceptionHandler.can_handle
08:18 dalek parrot: review: https://trac.parrot.org/parrot/changeset/40859/
08:20 iblechbot_ joined #parrot
08:23 dukeleto tests passing again on my linux box. i guess i forgot to do a reaclean
08:25 mikehh partcl r648 builds on parrot r40857 - make test - same 6 tests fail but all subtests PASS - Ubuntu 9.04 amd64 (g++)
08:27 mikehh cardinal builds on parrot r40857 - make test - 3 FAILures due to - get_pmc_keyed() not implemented in class 'String' - Ubuntu 9.04 amd64 (g++)
08:28 mikehh decnum_dynpmcs r181 builds on parrot r40857 - make test PASS - Ubuntu 9.04 amd64 (g++)
08:31 mikehh lua fails to build - error:imcc:make_pmc_const: no such pmc - in file 'lua.pir' line 307 - make: *** [lua.pbc] Error 1
08:32 mikehh bacek: ok - ready to go - any more commits comming soon?
08:32 bacek mikehh: not in next half-an-hour
08:32 mikehh bacek: ok here we go ....
08:32 * bacek crossing fingers
08:46 mikehh bacek: builds ok - make test -> packfile failures (others pass) - codetest failures
08:46 bacek mikehh: thanks! codetest failures left on purpose. packfile aren't critical atm.
08:49 MoC joined #parrot
08:51 mikehh bacek - I also get a backtrace from t/tools/parrot_debugger.t but reports ok - this was fixed a couple of days ago in trunk
08:51 bacek mikehh: O. Good to know.
08:52 Whiteknight joined #parrot
08:55 mikehh I have 3 windows with 16 tabs open in kansole at the moment
08:55 mikehh konsole
08:56 bacek mikehh: I've got few hundreds files open in vim :)
08:56 mikehh well not that many - but yeah :-}
08:58 mikehh i generally use Kate and gvim
09:00 bacek I'm using gvim as well.
09:01 dalek parrot: r40860 | bacek++ | branches/context_pmc3 (7 files):
09:01 dalek parrot: Move src/context.c into src/call/context.c and drop context.h
09:01 dalek parrot: review: https://trac.parrot.org/parrot/changeset/40860/
09:08 mikehh bacek: rakudo fails to build - p6invocation.c: In function ‘opcode_t* Parrot_P6Invocation_invoke(parrot_interp_t*, PMC*, void*)’: p6invocation.c:281: error: ‘struct PMC’ has no member named ‘lex_pad’
09:08 mikehh that's at r40859
09:11 bacek mikehh: yeah. branch is backward incompatible.
09:13 mikehh bacek: let me know when you want more testing on that branch done - I'll keep it in my dir
09:13 bacek mikehh: ok, thanks
09:28 dalek parrot: r40861 | bacek++ | branches/context_pmc3 (48 files):
09:28 dalek parrot: Replace 'cx' with 'pcc' prefix in context related functions
09:28 dalek parrot: review: https://trac.parrot.org/parrot/changeset/40861/
09:29 snarkyboojum joined #parrot
09:33 snarkyboojum joined #parrot
09:37 snarkyboojum joined #parrot
09:54 dalek parrot: r40862 | bacek++ | branches/context_pmc3 (4 files):
09:54 dalek parrot: [core] Add results_signature accessor to Context and use it
09:54 dalek parrot: review: https://trac.parrot.org/parrot/changeset/40862/
09:54 dalek parrot: r40863 | bacek++ | branches/context_pmc3 (4 files):
09:54 dalek parrot: Add Context flags accessors and use them
09:54 dalek parrot: review: https://trac.parrot.org/parrot/changeset/40863/
10:31 JimmyZ joined #parrot
10:31 iblechbot joined #parrot
11:10 bacek *incoming*
11:14 bacek_ joined #parrot
11:17 dalek parrot: r40864 | bacek++ | branches/context_pmc3 (2 files):
11:17 dalek parrot: [cage] Fix usage of param guards. Add Parrot_pcc_set_constants function.
11:18 dalek parrot: review: https://trac.parrot.org/parrot/changeset/40864/
11:18 dalek parrot: r40865 | bacek++ | branches/context_pmc3/src/embed.c:
11:18 dalek parrot: Use Parrot_pcc_flags functions instead of poking into struct directly.
11:18 dalek parrot: review: https://trac.parrot.org/parrot/changeset/40865/
11:18 dalek parrot: r40866 | bacek++ | branches/context_pmc3/src (3 files):
11:18 dalek parrot: Use Parrot_pcc_set_constants function.
11:18 dalek parrot: review: https://trac.parrot.org/parrot/changeset/40866/
11:18 dalek parrot: r40867 | bacek++ | branches/context_pmc3 (2 files):
11:18 dalek parrot: [cage] Add PARROT_CAN(NOT)_RETURN_NULL to context functions
11:18 dalek parrot: review: https://trac.parrot.org/parrot/changeset/40867/
11:18 dalek parrot: r40868 | bacek++ | branches/context_pmc3 (2 files):
11:18 dalek parrot: [cage] Drop redundant struct keyword
11:18 dalek parrot: review: https://trac.parrot.org/parrot/changeset/40868/
11:18 dalek parrot: r40869 | bacek++ | branches/context_pmc3 (3 files):
11:18 dalek parrot: Add Context handlers accessors and use them
11:18 dalek parrot: review: https://trac.parrot.org/parrot/changeset/40869/
11:18 dalek parrot: r40870 | bacek++ | branches/context_pmc3/t/native_pbc (4 files):
11:18 dalek parrot: Rebuild native pbc
11:18 dalek parrot: review: https://trac.parrot.org/parrot/changeset/40870/
11:21 dalek parrot: r40871 | bacek++ | branches/context_pmc3 (2 files):
11:21 dalek parrot: Steal functions from first commit to context_pmc2 branch for accessing registers. Whiteknight++
11:21 dalek parrot: review: https://trac.parrot.org/parrot/changeset/40871/
11:21 dalek parrot: r40872 | bacek++ | branches/context_pmc3 (2 files):
11:21 dalek parrot: Use more Context accessors functions
11:21 dalek parrot: review: https://trac.parrot.org/parrot/changeset/40872/
11:25 mikehh_ joined #parrot
11:39 kid51 joined #parrot
11:49 snarkyboojum joined #parrot
11:51 snarkyboojum joined #parrot
12:12 jan joined #parrot
12:37 dalek parrot: r40873 | bacek++ | branches/context_pmc3 (5 files):
12:37 dalek parrot: Rename Parrot_pcc_get_context to Parrot_pcc_get_context_struct
12:37 dalek parrot: review: https://trac.parrot.org/parrot/changeset/40873/
12:46 HG` joined #parrot
12:58 mikehh joined #parrot
13:04 dalek parrot: r40874 | bacek++ | branches/context_pmc3/src/gc/alloc_register.c:
13:04 dalek parrot: [cage] Add more docs.
13:04 dalek parrot: review: https://trac.parrot.org/parrot/changeset/40874/
13:04 dalek parrot: r40875 | bacek++ | branches/context_pmc3/src/pmc/sub.pmc:
13:04 dalek parrot: [cage] Fix codetest failure in Sub PMC.
13:04 dalek parrot: review: https://trac.parrot.org/parrot/changeset/40875/
13:04 dalek parrot: r40876 | bacek++ | branches/context_pmc3/src/hash.c:
13:04 dalek parrot: [cage] Fix codetest failure in hash.c
13:04 dalek parrot: review: https://trac.parrot.org/parrot/changeset/40876/
13:04 dalek parrot: r40877 | bacek++ | branches/context_pmc3/include/parrot/warnings.h:
13:04 dalek parrot: [cage] Fix codetest failures.
13:04 dalek parrot: review: https://trac.parrot.org/parrot/changeset/40877/
13:04 dalek parrot: r40878 | bacek++ | branches/context_pmc3 (3 files):
13:04 dalek parrot: Add Parrot_pcc_get_(int|num)_constants functions and use it.
13:04 dalek parrot: review: https://trac.parrot.org/parrot/changeset/40878/
13:06 bacek_ mikehh: Next round of testing?
13:07 bacek_ see you tomorrow?
13:40 mikehh msg bacek post_config test failures, make test has t/dynpmc and t/dynoplibs failures and codetest failures
13:40 purl Message for bacek stored.
14:05 payload joined #parrot
14:09 silug joined #parrot
14:19 Psyche^ joined #parrot
14:30 snarkyboojum joined #parrot
14:36 ruoso joined #parrot
14:43 mokurai joined #parrot
14:55 cognominal joined #parrot
14:58 bacek joined #parrot
15:07 JimmyZ joined #parrot
15:16 jrtayloriv Anyone know of any tools that can generate a graph of include dependencies for parrot? None of the scripts I've found in Google are working. BTW -- this would be a very nice thing to have in docs/ for new developers (could possibly be autogenerated at build)
15:18 mikehh I don't know about a graph but there is some documentation around - let me check
15:20 jrtayloriv It would be really nice if Configure.pl could check if graphviz is installed, and if so generate a dependency graph during make
15:24 mikehh I agree - the only documemntation I could find is in docs/book/aft/appd_build_options.pod- used to be part of the html docs but not anymore
15:25 mikehh sorry my wireless keyboard is dropping keys - thats docs/book/draft/appd_build_options.pod
15:26 mikehh should be somewhere else but can't find it at the moment
15:27 jrtayloriv mikehh, oh, sorry -- I mean header file dependencies (like what #include's what)
15:28 cognominal joined #parrot
15:28 mikehh ah ha - see what youi mean - and no - not that I have seen - maybe around but I haven't noticed
15:39 iblechbot joined #parrot
15:41 dalek TT #959 created by jrtayloriv++: Auto-generate #include dependency graphs in docs/ at build.
15:41 JimmyZ joined #parrot
16:00 mberends joined #parrot
16:08 Coke these segfaults are killing me.
16:14 mikehh Coke: wait for the next installment of branches :-}
16:21 dalek TT #960 created by coke++: segfaults in Parrot_String_mark
16:24 Coke mikehh: you joke, but I wonder how HLL authors who are not on the board view all these segfaults. :|
16:25 mikehh actually in the context_pmc3 branch partcl is the only HLL that seems to give the same results
16:26 mikehh rakudo doesn't even build and cardinal fails most tests
16:30 mikehh Coke: that's on partcl make test - I haven't even attempted make spectest
16:33 mikehh Coke: as I see it most of the segfaults seem to come from some aspect of gc
16:34 mikehh maybe we need a complete re-write not just a refactoring of gc
16:35 petdance joined #parrot
16:36 mikehh petdance: you seem to be here as both Andy and petdance :-}
16:37 petdance I'm prob'ly still on at Work.
16:37 mikehh petdance: how's your gc knowledge
16:38 petdance nil
16:39 mikehh one of the most disasterous projects I was ever involved with had a large gc component
17:07 petdance joined #parrot
17:30 MoC joined #parrot
17:36 rbaumer joined #parrot
17:44 Zak joined #parrot
17:57 mikehh_ joined #parrot
18:05 mikehh joined #parrot
18:08 kjeldahl joined #parrot
18:33 Coke (disaster, large gc component) Oh, so, parrot, then? =-)
18:43 HG` joined #parrot
18:43 * Coke ponders writing "A plea from partcl-land.", in the style of pmichaud.
18:44 dukeleto that would be interesting
18:44 * Coke wonders if rakudo is seeing an increase in segfaults recently or if it's just him.
18:44 Coke s/him/partcl/
18:45 dukeleto Coke: do you think the hllmacros could be to blame?
18:46 joeri joined #parrot
18:46 moritz Coke: I'm about half way through rakudo's spectest with newest parrot, no segfaults so far
18:46 Coke moritz: ah well.
18:47 Coke dukeleto: I wouldn't think so, but who knows.
18:50 payload joined #parrot
18:52 Coke msg cotto - does the profiler work on dynamically loaded ops?
18:52 purl Message for cotto stored.
18:53 dukeleto it's funny how small proof-of-concept scripts quickly grow to be behemoth's
18:54 jrtayloriv Would the macro PARROT_CONTINUATION() be usable in any other .pmc files that use the Continuation PMC? Should I use PARROT_CONTINUATION in other PMCs, or GET_ATTR_Continuation, or does it not matter?
19:02 ZeroForce joined #parrot
19:05 dukeleto good localtime()
19:38 Coke jrtayloriv: that macro is only defined, never used; looking at what it does, I'd recommend using the attr macros instead.
19:40 Coke jrtayloriv: in fact, I don't think that the pmc has a 'cont' accessor anymore.
19:40 jrtayloriv Coke, OK -- is that just a special case for this PMC, or for all PMC's? Because docs/pmc.pod says: "For given pmc ("Foo"), an ATTR ("bar") can be accessed either directly: "PARROT_FOO(pmc)->bar" or via a SETATTR/GETATTR accessor macro: "GETATTR_Foo_bar(INTERP, x)" ... should I update the docs to remove the reference to the PARROT_FOO macro?
19:40 Coke (so at first reading, looks like that macro is doomed to fail.)
19:42 Tene Has the issue of Parrot's ssl cert expiring already been discussed?
19:42 Coke I would recommend striking that from the docs, yah; just use the ATTR macros; but: I didn't write that subsystem nor am I a c programmer. =-)
19:42 Coke Tene: there's a ticket.
19:42 * Tene nods.
19:44 jrtayloriv Coke, OK -- thank you.
19:45 Coke thanks for going through docs and not just giving up out of frustration. =-)
19:46 * Coke wonders what the point of macros like gc_flag_CLEAR is.
19:46 Coke (they're just macro aliases.)
19:49 Coke pir question - when invoking a tcl .sub, I have to jump through hoops, due to tcl's autoloading feature. autoload itself (called unknown in tcl) is a user visible sub that can be moved out the way, so the logic is:
19:49 Coke find actual sub; failing that, find a global named 'unknown' and invoke that (it does magic to load the sub from several possible locations); failing finding that sub, emit an error about an invalid sub name.
19:50 Coke I'd like to wrap all that up into an opcode, or perhaps a method on the interpreter.
19:50 Coke I don't think I can write an opcode that automagically handles variable args. Any ideas?
19:50 dukeleto coke: sounds like a good place for a dynop
19:51 dukeleto coke: the possibility of varargs is unknown to me. possibly?
19:51 Coke dukeleto: pretty sure no; I'd need to either take a single slurpy PMC or have N variants. (which isn't feasible)
19:51 Coke I hesitate to create YA PMC.
19:52 Coke (so I'm leaning towards method that I glom onto the parrot interpreter object.)
19:54 Coke (but I hesitate to invoke something with PCC more than I have to. =-)
19:54 Coke I remember a time, long long ago, when the PCC were going to be optional.
19:54 ZeroForce joined #parrot
19:56 ZeroForce joined #parrot
19:58 Coke ZeroForce: hi!
19:58 ZeroForce Coke: Hey there :)
19:59 ZeroForce Congrats everyone on the releases this year!
20:05 asdf joined #parrot
20:09 ZeroForce1 joined #parrot
20:11 cotto Coke, the profiler shouldn't have trouble with dynops.
20:12 dalek parrot: r40879 | dukeleto++ | trunk/tools/dev/parrot_shell.pl:
20:12 dalek parrot: [tools] The Parrot Shell is a rapid prototyping tool for writing Parrot code
20:12 dalek parrot: review: https://trac.parrot.org/parrot/changeset/40879/
20:13 Coke dukeleto: "its own interpreter"
20:14 dukeleto coke: patches welcome ;)
20:14 Coke cotto: I'm not seeing any of my tcl_ ops in your run.
20:14 Coke could it be mislabeling them?
20:15 nopaste "jrtayloriv" at 96.238.199.12 pasted "[PATCH]: fix docs/pmc.pod" (18 lines) at http://nopaste.snit.ch/17745
20:15 cotto I'll look into it later today when I have a minute.
20:16 jrtayloriv Coke: That look OK to you, before I post it up on trac?
20:16 cotto Do you have an example that doesn't work, preferably simple?
20:16 payload joined #parrot
20:23 mokurai left #parrot
20:26 mokurai joined #parrot
20:26 Coke jrtayloriv: seems reasonable to me.
20:26 Coke cotto: danke
20:31 joeri left #parrot
20:32 bacek joined #parrot
20:39 dalek parrot: r40880 | mikehh++ | trunk/MANIFEST:
20:39 dalek parrot: fix manifest_tests failure
20:39 dalek parrot: review: https://trac.parrot.org/parrot/changeset/40880/
20:39 chromatic joined #parrot
20:42 cotto Coke, did that alternate patch work for you?
20:44 ZeroForce joined #parrot
20:45 cotto stupid hack incoming.  alternatives welcome.
20:46 dalek parrot: r40881 | cotto++ | branches/pluggable_runcore/src/runcore/cores.c:
20:46 dalek parrot: [profile] mark the runcore->profile_filename STRING during each runloop iteration.
20:46 dalek parrot: This is stupid and bad and wrong, but the GC doesn't catch it by default and
20:46 dalek parrot: apparently marking it when a runloop starts isn't always sufficient.
20:46 dalek parrot: review: https://trac.parrot.org/parrot/changeset/40881/
20:47 dukeleto mikehh++
20:48 * dukeleto beats the GC for being naughty
20:48 chromatic cotto, we need to figure out why it isn't getting marked.
20:48 chromatic If we need to use pmc_register or whatever it is now, that's fine too.
20:51 cotto I suspect that it's not getting marked because profiling_runcore_t is larger than runcore_t, but that's just a guess.
20:52 cotto (and the STRING* is after the end of the runcore_t part of the struct)
20:56 dalek parrot: r40882 | dukeleto++ | trunk/tools/dev/parrot_shell.pl:
20:56 dalek parrot: [cage] Fix svn properties of tools/dev/parrot_shell.pl and add moar grammerz, coke++
20:56 dalek parrot: review: https://trac.parrot.org/parrot/changeset/40882/
20:56 bacek Good morning
20:56 purl And good moroning to you, bacek.
20:58 cotto Coke, the profiling runcore doesn't have any trouble with the dynops in t/dynoplibs/obscure.t or math.t .
21:00 cotto I'd appreciate an example of how it goes south for you.
21:00 dduncan joined #parrot
21:03 chromatic Runcores don't get marked.  Hm.
21:09 rjbs joined #parrot
21:09 rjbs I tried to view http://trac.parrot.org/parrot/wiki/ModuleEcosystem
21:09 rjbs I was redirected to that URL at http.  The cert is not trusted by Firefox, which now wants me to click about 900 things to be allowed to look at the page.
21:09 rjbs Does that need to be done over SSL?  It would be nice if I didn't need an exception to my security policy just to read the wiki.
21:10 dalek rakudo: 80fdc48 | last.of.the.careless.men@gmail.com++ | src/setting/Complex.pm:
21:10 dalek rakudo: Implement Complex version of **.
21:10 dalek rakudo: Signed-off-by: Moritz Lenz <moritz@faui2k3.org>
21:10 dalek rakudo: review: http://github.com/rakudo/rakudo/commit/8​0fdc4835ec664cd613cf24c3444ca5a0f550653
21:10 moritz it would be nice if we had a non-expired SSL cert
21:10 cotto Coke, msg me with a nopaste example.  I'll be back tonight.
21:12 rjbs left #parrot
21:17 nopaste "NotFound" at 213.97.96.43 pasted "Patch: avoid partcl 'attempt to access code outside of current code segment' errors in make test" (32 lines) at http://nopaste.snit.ch/17747
21:22 NotFound Coke: can you look at this?
21:37 Whiteknight joined #parrot
21:41 bacek Whiteknight: ping
21:41 Whiteknight bacek: pong
21:42 bacek Whiteknight: do you have any chance to look into src/jit/i386/jit_defs.c, line 875? (context_pmc3)
21:42 Whiteknight a little bit, yes
21:43 bacek There is commented-out line. And I have no idea how to fix it.
21:46 allison joined #parrot
21:49 dukeleto joined #parrot
21:51 rbaumer joined #parrot
22:04 Whiteknight bacek: Okay, that's tricky but I think I can figure out how to fix it
22:04 bacek Whiteknight: sigh...
22:05 Whiteknight actually, I don't think that line of code does anything
22:05 chromatic gc_register_pmc and gc_unregister_pmc could just as easily take PObjs.
22:05 chromatic STRINGs and PMCs both.
22:06 bacek Whiteknight: I have no idea about i386 assembly and jit in parrot...
22:07 Whiteknight bacek: The whole system is a tragedy
22:08 Whiteknight what JIT in Parrot does is slow down the pace of development, because we have to maintain it when we make changes, and it doesn't really provide any speed improvements
22:08 bacek Whiteknight: I heard rumours about it. But I can't just ignore jit...
22:08 Whiteknight so it's really a useful thing
22:09 chromatic That's not entirely true.  Auto-generated NCI thunks are useful.
22:09 Whiteknight chromatic: true, but that's a very small portion of the codebase that is the JIT system
22:09 chromatic True.
22:10 Whiteknight and I would suggest we could develop a much better, more robust, more flexible, and more platform-independent frame builder without the current stuff
22:10 Whiteknight the current JIT system is little more then a roadblock to developing better things
22:12 chromatic Maybe we should consider deprecating the current JIT at 2.0.
22:13 bacek Why it require deprecation notice?
22:14 moritz because it's part of some kind of API
22:14 moritz like parrot --runcore=jit
22:14 moritz or so
22:14 moritz that's documented, so it has to be deprecated before removal
22:15 Whiteknight chromatic: I would like that very much, but I'm a known proponent of removing bad stuff to make way for better stuff
22:16 Whiteknight I'm a big believer that often it's easier to just start from scratch
22:16 chromatic Counterexample: PCC rewiring.
22:16 moritz I tend to think that this kind of thinking underestimates the effort put in the old system
22:18 Whiteknight chromatic: what do you mean by that?
22:18 chromatic We're on attempt #4 or #5 to rewrite an old system.
22:19 dalek parrot: r40883 | mikehh++ | trunk/tools/dev/parrot_shell.pl:
22:19 dalek parrot: fix cuddled else(s) in tools/dev/parrot_shell.pl
22:19 dalek parrot: review: https://trac.parrot.org/parrot/changeset/40883/
22:20 mikehh Whiteknight: +1
22:20 purl 1
22:20 Whiteknight chromatic: and absolutely no progress was being made on refactoring it in place
22:21 chromatic I don't believe that.
22:21 Whiteknight at least in these attempted rewrites progress is being made
22:21 chromatic At least in Allison's first attempt, she had *all tests passing* at one point.
22:21 mikehh I'm still gettinga perlcritic failure on tools/dev/parrot_shell.pl - Missing properly located perl coda for parrot source at /home/mhh/x.parrot/tools/dev/parrot_shell.pl line 159
22:22 moritz well, I don't see a coda in tehre
22:22 mikehh perlcritic test 257
22:23 preflex joined #parrot
22:23 moritz $ tail -n 6  tools/dev/symlink.pl
22:23 moritz that's how it should look
22:25 dalek rakudo: 9fff158 | pmichaud++ | src/setting/Rat.pm:
22:25 dalek rakudo: First-draft implementation of Rat datatype.
22:25 dalek rakudo: review: http://github.com/rakudo/rakudo/commit/9​fff15880d11473fc480b6c167eca0df8bd5b02b
22:25 mikehh moritz: ok - I didn't undesand the error - will fix now
22:26 Whiteknight chromatic: what's going on in that branch right now?
22:26 chromatic Test fixes.
22:29 Whiteknight how many tests failing?
22:30 chromatic Some 1500, last I heard.
22:31 moritz Whiteknight: PGE doesn't build
22:31 Whiteknight oh wow, more then I last remember hearing
22:31 moritz I'm running coretest right now
22:31 dalek rakudo: cfb7d8d | pmichaud++ | build/Makefile.in:
22:31 dalek rakudo: Forgot to commit updated Makefile.
22:31 dalek rakudo: review: http://github.com/rakudo/rakudo/commit/c​fb7d8db3baf784ed0ebb775ef9f56f2917a64ce
22:31 Whiteknight moritz: me too
22:33 dalek parrot: r40884 | mikehh++ | trunk/tools/dev/parrot_shell.pl:
22:33 dalek parrot: fix perlcritic code error in tools/dev/parrot_shell.pl
22:33 dalek parrot: review: https://trac.parrot.org/parrot/changeset/40884/
22:33 moritz it seems to hang at some point
22:33 mikehh bah I ment to type coda not code
22:35 moritz in t/pmc/capture.t it hangs (pcc branch)
22:36 moritz bed time for me, TTFN folk
22:36 bacek What "-g" command line arg do?
22:36 bacek moritz: good night
22:41 rg joined #parrot
22:42 mikehh bacek: did you get my message about the test failures in the context_pmc3 branch
22:43 bacek mikehh: yes, thanks.
22:43 bacek What is "post_config" test?
22:44 mikehh running with --test on perl Configure.pm - after the configure
22:44 chromatic It tests that the configuration step performed as expected.
22:44 bacek hm...
22:45 bacek It passed on my box.
22:46 bacek Ah. Found it
22:48 jrtayloriv If a PMC "Bar" extends another PMC "Foo", and Foo has ATTRs, does "Bar" have Foo's ATTRs in addition to whatever it declares itself? i.e. If I can say PARROT_FOO(foo_pmc)->foos_attr, and Bar extends Foo, can I say PARROT_BAR(bar_pmc)->foos_attr ?
22:48 mikehh pre is before, but it won't configure if it fails that - they are actually --test=configure and --test=build
22:48 Whiteknight pcc_args_unify builds PGE fine on my system
22:49 Whiteknight doesn't build TGE though
22:49 jrtayloriv nm -- I found the answer
22:50 bacek mikehh: it passed after realclean on my box. Can you nopaste your results?
23:01 bacek $dayjob time
23:01 bacek see you!
23:16 Coke NotFound: holy crap it works.
23:17 Coke NotFound++
23:17 Coke NotFound++
23:17 Coke NotFound++
23:17 Coke NotFound++
23:18 Coke # of course, I have no idea why 'end' craps out when 'exit 0' works fine, but nice.
23:18 davidfetter heh
23:18 Coke parrot can deal with the original bug report, I'll use the workaround. =-)
23:19 Coke ah. it avoid that error, but a wholesale swap out of end for 'exit FOO' causes other test failures. checking...
23:20 dduncan left #parrot
23:24 Coke ... because I'm assuming that tcl always exits with 0. Making that explicit for now, whee.
23:28 dalek rakudo: 0712c33 | pmichaud++ | src/setting/ (2 files):
23:28 dalek rakudo: Move Int-based operators into src/setting/Int.pm .
23:28 dalek rakudo: review: http://github.com/rakudo/rakudo/commit/0​712c333bc0ebb73aa93e308517eb2e97cec31f0
23:30 kthakore joined #parrot
23:30 kthakore chromatic, can I ask you a question on #sdl ?
23:30 chromatic Yes, if it's quick.
23:30 kthakore its about spliting SDL.xs
23:34 Coke NotFound++
23:37 Coke (now I can do "make test && git commit -a" ! =-)
23:37 dalek partcl: r649 | coke++ | wiki/ParrotIssues.wiki:
23:37 dalek partcl: NotFound++ # we can avoid this bug!
23:37 dalek partcl: review: http://code.google.com/p/p​artcl/source/detail?r=649
23:37 dalek partcl: r650 | coke++ | trunk/src/tclsh.pir:
23:37 dalek partcl: Avoid parrot RT#57088; 'make test' now runs clean.
23:37 dalek partcl: NotFound++ for the workaround.
23:37 dalek partcl: Note: we are still leaving parrot with the wrong exit value in
23:37 dalek partcl: some cases; we're just doing it explicitly now.
23:37 dalek partcl: review: http://code.google.com/p/p​artcl/source/detail?r=650
23:43 nopaste "mikehh" at 90.209.128.129 pasted "post-config and make test failures with context_pmc3 branch for bacek" (163 lines) at http://nopaste.snit.ch/17753
23:43 Coke cottyou in?
23:43 Coke er, cotto?
23:45 NotFound Coke: do you like my little trick? ;)
23:45 Coke absolutely. Thanks!
23:47 Coke msg cotto apparently I just didn't happen to hit a dynop; if I run profiling on "tclsh -e 'break'", I see it enter the break sub, and then invoke the tcl_break dynop. I would have expected to see one before, but must have been wrong; will let you know if I have any more problems.
23:47 purl Message for cotto stored.
23:48 Coke hurm. I wonder if I should provide a way to run tclsh without running library/init.tcl first.
23:48 Coke (guessing no, but damn would it be nice for speed. :|)
23:49 mikehh All tests PASS (pre/post-config, smoke, nqp_test, fulltest) at r40884 - Ubuntu 9.04 amd64 (g++)
23:50 mikehh more karma for g to come :-}
23:54 Coke oh, or we could just make parrot faster!
23:55 Coke If anyone is interested in helping me come up with a way to make partcl avoid doing runtime dispatch all the time, I'd love the help. =-)
23:56 Coke (mainly because doing the runtime dispatch instead a loop ends up recompiling a lot of stuff we shouldn't have to recompile.)
23:56 mikehh partcl r650 builds on parrot r40884 - make test - PASS - Ubuntu 9.04 amd64 (g++)
23:57 dalek rakudo: 8772e90 | pmichaud++ | src/setting/Int.pm:
23:57 dalek rakudo: Some more conversions (Int.Num, Int.Rat, Int.Int).
23:57 dalek rakudo: review: http://github.com/rakudo/rakudo/commit/8​772e906286f4375e1847a52c4cd57e8aa2032df
23:57 mikehh Coke: make test PASS NotFound++

| Channels | #parrot index | Today | | Search | Google Search | Plain-Text | summary

Parrot | source cross referenced