Camelia, the Perl 6 bug

IRC log for #parrot, 2012-05-04

Parrot | source cross referenced

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

All times shown according to UTC.

Time Nick Message
00:00 whiteknight bacek_at_work: I don't think it's a problem
00:00 bacek_at_work whiteknight, which one?
00:04 whiteknight aloha saying karma values
00:04 benabik whiteknight: You appear to be in the minority.
00:05 benabik I was thrown off by it, but there seems to be active vitriol towards it elsewhere.
00:06 whiteknight where, in #perl6?
00:06 benabik I seem to recall some complaints here, but definitely there, yes
00:06 Coke I'd rather not have it announced, myself.
00:13 whiteknight that's fine too. I won't argue about it
00:32 nbrown joined #parrot
00:34 cotto ~~
00:38 nbrown hey cotto, I saw you were looking for me
00:45 cotto nbrown, yup.  been thinking about m0's memory model
00:45 cotto I might be making it out to be more complex than necessary.  lemme find my gist
00:46 cotto https://raw.github.com/gist/2581153/8e546ebf1​94eaa858059e27f987c058f5e1b0e90/questions.txt
00:46 nbrown glad someone has been. I got buried this week and am slowly finding my way ou
00:46 cotto no worries
00:47 nbrown first thing, I don't think there's any point to having S type constants, do you?
00:48 cotto they're shorthand for p
00:49 cotto we could have INPP
00:49 cotto or have then not be typed at all
00:49 cotto i.e. Rn could be any of insp
00:49 nbrown oops, I really meant P, not S
00:50 cotto P will be useful once we have a meaningful way of representing objects
00:50 nbrown so P would be for a serialized object?
00:50 cotto it'd be an active object
00:50 cotto well, a pointer to one
00:51 nbrown but isn't a constants section read out of a m0b file?
00:51 cotto they have a runtime meaning too
00:51 nbrown m0 doesn't really have objects
00:51 cotto or will
00:51 nbrown I don't think I understand the runtime meaning of the constants section
00:52 nbrown *segment
00:52 nbrown I thought the segment was simply a representation of the constants in the m0b file
00:52 cotto yes, that's the role of the constants segment
00:53 nbrown ok. I was getting mildly confused between the constants segment and the data segment
00:53 cotto SP registers can point at entries in the constants table or at dynamic objects in the data segment
00:55 nbrown right, I'm getting the hang of it now. I was misinterpreting things
00:56 Khisanth joined #parrot
00:58 nbrown btw, I really want to write an m0 debugger, I just haven't been sure how to do it yet
00:58 cotto I'd love you forever.
00:59 nbrown Everytime I've had to debug an m0 test file I've wanted one oh so badly
00:59 cotto you and me both
01:00 nbrown :)
01:00 cotto I wonder if named registers would be a bad idea.
01:01 nbrown so on the recursion thing, where would we store the state if we clone after entering? another callframe?
01:01 nbrown I understand the appeal of named registers, but I see them just complicating anything that wants to generate m0
01:02 cotto the current callframe's PCF points at the callframe that called it
01:02 cotto (need to verify, but the poke_caller test does this)
01:03 nbrown the poke_caller test populates the new callframe prior to entering
01:03 cotto ah.  I see the question
01:04 nbrown I was trying to understand what you meant when you said leftover state
01:05 cotto actually, I can see it working like it does now
01:06 kid51_ joined #parrot
01:06 nbrown ok, I think the current approach works pretty well
01:09 nbrown so on the unresolved questions, if the global data is purely runtime, how do we access it?
01:10 nbrown and how much gets allocated?
01:12 cotto it'd fit nicely with the other global interp info
01:12 cotto CONFIG, ARGV, ARGC, etc
01:13 nbrown right, but those are all known at load time
01:13 nbrown I guess we can deal with it the same way we deal with the data segment
01:14 nbrown fixed size or grow as needed
01:14 cotto yes
01:14 nbrown ok
01:15 nbrown Your gist seems pretty on point to me
01:16 nbrown I don't think you're making it too comlpex
01:16 cotto how about sbrk?
01:16 cotto and implementing virtual memory?
01:16 nbrown I don't know enough about it
01:16 kid51 joined #parrot
01:16 cotto (either that or realloc)
01:17 nbrown I would have used realloc since I know that
01:17 nbrown I don't see a way around it if we allow the data segment and global data to grow
01:17 cotto yes
01:18 cotto if realloc doesn't move memory, it'll work nicely
01:18 nbrown I believe it may
01:18 cotto realloc would at least be a good first approximation
01:19 nbrown I just looked it up and it says that it can move the memory block to a new location
01:19 nbrown I'll go read up on sbrk
01:21 cotto sbrk is what malloc is implemented on top of
01:21 cotto basically increases the amount of memory available to a process
01:21 nbrown ok
01:22 nbrown makes sense to me
01:23 nbrown so which part were you concerned you werw overthinking?
01:23 cotto implementing a virtual memory system
01:23 cotto if I get to that point, it makes me think that something is wrong
01:26 nbrown fair enough, so if we step back: can we just have an array of data sections?
01:26 nbrown use the malloc / gc_alloc ops to generate new data segments?
01:27 cotto possible
01:27 cotto interesting idea
01:27 nbrown it's far simpler
01:28 cotto I'm not sure about it being *far* simpler than just using realloc to expand the data segment, but the idea definitely is worth considering
01:29 nbrown yeah, as I'm thinking about it *far* is *far* too strong a word
01:29 nbrown :)
01:29 nbrown but it would allow a faster prototype to play with
01:30 nbrown scratch that, I meant allow us to write a prototype faster
01:31 cotto so how would you point to byte 32 in data segment 4?
01:32 nbrown so the part that was starting to trip me up was the array of pointers to data segments....
01:32 nbrown is it something that's fixed size?
01:33 cotto I was jus tthinking about that
01:33 cotto there'd need to be a way to expand it
01:34 cotto expanding it implicitly is an option, but that's also slightly magical
01:34 nbrown we'd need something like a directory segment to point data segments, but that's difficult to grow
01:36 cotto it could look like cf[DATA][n]
01:37 cotto realloc for that would probably be workable, but that's one of the side-effects that M0 should avoid if at all possible
01:37 nbrown yeah
01:38 nbrown if we preallocate space for a decent number of data segment pointers, it's likely we could avoid it
01:39 nbrown maybe we could even allow the user to specify the # of desired data segments in the m0 file to help avoid the realloc case
01:39 nbrown something like .num_data_segments 20
01:39 nbrown or whatever
01:40 cotto I'm still not quite sold on it.
01:40 nbrown and if the user didn't provide anything m0 has some reasonable default
01:40 nbrown me neither
01:40 cotto flat memory is very non-magical
01:41 nbrown flat memory?
01:41 cotto a single data segment that needs to be manually expanded and shrunk
01:42 nbrown I don't think the array of data segments needs to be magical
01:43 dalek rakudo/nom: 7eae9b2 | pmichaud++ | src/ (2 files):
01:43 dalek rakudo/nom: Fix [|] and [&] to provide flat junctions (RT #112696, masak++).
01:43 dalek rakudo/nom: review: https://github.com/rakudo/rakudo/commit/7eae9b2740
01:43 cotto a better word would be "simple"
01:43 nbrown yes, I'd agree with that
01:44 nbrown generating code for multiple data segments would be more complicated
01:49 nbrown hmmm, I'm coming back around to the virtual memory idea
01:51 nbrown accessing/writing data in multiple data segments would be not a lot of fun
02:06 dalek rakudo/nom: 2d2584a | pmichaud++ | src/core/metaops.pm:
02:06 dalek rakudo/nom: Implement triangle forms for junction reducers:   [\|]  and [\&]
02:06 dalek rakudo/nom: review: https://github.com/rakudo/rakudo/commit/2d2584a625
03:33 rurban_home https://github.com/parrot/parrot/pull/762
03:38 aloha (parrot/parrot) Issues opened : 762 (mark stack-overflow gc for clang -faddress-sanitizer) by rurban : https://github.com/parrot/parrot/issues/762
03:52 dalek parrot: 7433886 | rurban++ | src/gc/system.c:
03:52 dalek parrot: mark stack-overflow  gc for clang -faddress-sanitizer
03:52 dalek parrot:
03:52 dalek parrot: http://clang.llvm.org/docs/AddressSanitizer.html fails in
03:52 dalek parrot: trace_mem_block() with a stack invalid READ, which is actually valid in our GC
03:52 dalek parrot: . Since this is the only function we just use the recommended detection and attribute
03:52 dalek parrot: here once.
03:52 dalek parrot: review: https://github.com/parrot/parrot/commit/7433886ded
03:52 dalek parrot: 68a170c | jimmy++ | src/gc/system.c:
03:53 dalek parrot: Merge pull request #762 from rurban/master
03:53 dalek parrot:
03:53 dalek parrot: mark stack-overflow gc for clang -faddress-sanitizer
03:53 dalek parrot: review: https://github.com/parrot/parrot/commit/68a170c4e4
03:53 aloha (parrot/parrot) Issues closed : 762 (mark stack-overflow gc for clang -faddress-sanitizer) by rurban : https://github.com/parrot/parrot/issues/762
03:58 ttbot Parrot 68a170c4 i386-linux-thread-multi make error http://tt.taptinder.org/cmdinfo/81818
04:05 travis-ci joined #parrot
04:05 travis-ci [travis-ci] parrot/parrot#284 (master - 68a170c : Jimmy Zhuo): The build was broken.
04:05 travis-ci [travis-ci] Change view : https://github.com/parrot/par​rot/compare/b58a50c...68a170c
04:05 travis-ci [travis-ci] Build details : http://travis-ci.org/parrot/parrot/builds/1242128
04:05 travis-ci left #parrot
04:22 alester joined #parrot
04:22 alester hey rurban the 7433886ded3df005351e9a0c087ef674771b3488 change to src/gc/system.c breaks my GCC build
04:24 benabik It broke gcc and g++ on travis as well
04:24 alester I think we ought to flag attributes using the same technique in include/parrot/compiler.h for PARROT_PURE_FUNCTION and the like.
04:24 alester so that the detail of __attribute__((no_address_safety_analysis)) can be hidden in a macro
04:25 alester I'm watching a movie for the next half hour if you'd like to discuss.
04:29 benabik Given that we might want to handle the flag differently and that it breaks gcc/g++, perhaps we should revert the change?
04:31 alester I don't have a strong feeling either way.
04:31 alester But I'm writing to parrot-dev right now how I suggest we do it going forward.
04:32 benabik alester++
04:36 alester ok sent
04:55 rurban joined #parrot
05:51 contingencyplan_ joined #parrot
06:01 woosley joined #parrot
06:04 cotto ~~
06:33 fperrad joined #parrot
06:59 brrt joined #parrot
07:08 woosley joined #parrot
07:26 plobsing joined #parrot
07:36 preflex joined #parrot
08:05 lucian joined #parrot
08:10 moritz \o
08:10 brrt \o
08:11 nopaste joined #parrot
08:32 woosley joined #parrot
08:39 nopaste joined #parrot
08:43 nopaste joined #parrot
09:36 bacek_at_work jfyi about C99 in VC++ http://herbsutter.com/2012/05/03/re​ader-qa-what-about-vc-and-c99/?nope
09:36 bacek_at_work answer is "no"
09:36 bacek_at_work long answer is "We recommend that C developers use the C++ compiler to compile C code (using /TP if the file is named something.c). This is the best choice for using Visual C++ to compile C code."
09:37 moritz meh.
10:26 plobsing joined #parrot
10:40 dalek parrot: 1eff807 | NotFound++ | src/gc/system.c:
10:40 dalek parrot: fix check for address_sanitizer feature
10:40 dalek parrot: review: https://github.com/parrot/parrot/commit/1eff8070f0
10:58 travis-ci joined #parrot
10:58 travis-ci [travis-ci] parrot/parrot#285 (master - 1eff807 : NotFound): The build was fixed.
10:58 travis-ci [travis-ci] Change view : https://github.com/parrot/par​rot/compare/68a170c...1eff807
10:58 travis-ci [travis-ci] Build details : http://travis-ci.org/parrot/parrot/builds/1243696
10:58 travis-ci left #parrot
11:03 JimmyZ joined #parrot
11:57 dalek rakudo/nom: d337e35 | moritz++ | src/Perl6/ (2 files):
11:57 dalek rakudo/nom: carry around the name of the to-be-imported module, and use it in an error message
11:57 dalek rakudo/nom: review: https://github.com/rakudo/rakudo/commit/d337e35ac1
12:01 whiteknight joined #parrot
12:11 whiteknight good morning, #parrot
12:12 moritz \o whiteknight
12:12 whiteknight hello moritz
12:28 JimmyZ joined #parrot
13:01 PacoAir joined #parrot
13:10 dalek parrot: 5810fe5 | petdance++ | / (3 files):
13:10 dalek parrot: Added the PARROT_NO_ADDRESS_SAFETY_ANALYSIS macro, plus support for it in the Headerizer.
13:10 dalek parrot: review: https://github.com/parrot/parrot/commit/5810fe50c4
13:13 whiteknight petdance++ indeed
13:14 Coke petdance++
13:16 whiteknight and rurban++ for coming up with the idea
13:37 JimmyZ_ joined #parrot
13:43 bluescreen joined #parrot
13:56 jashwanth joined #parrot
14:04 alester joined #parrot
14:17 davidfetter joined #parrot
14:19 benabik joined #parrot
14:31 alester ping rurban
14:36 rurban here
14:37 rurban I see, you added a macro for asan
14:49 rurban Parrot_api_string_export_wchar writes one byte too much
14:49 rurban It should be wstrout = (wchar_t *) malloc(sizeof (wchar_t) * (len + 1));
14:52 jashwanth joined #parrot
14:52 Infinoid joined #parrot
14:52 Tene joined #parrot
14:52 jlaire joined #parrot
14:52 ttbot joined #parrot
14:52 sri joined #parrot
14:52 eternaleye joined #parrot
14:52 mdupont joined #parrot
14:52 schm00ster joined #parrot
14:52 dngor joined #parrot
14:52 dukeleto joined #parrot
14:52 bacek_at_work joined #parrot
14:52 perlite joined #parrot
14:52 aloha joined #parrot
14:52 Patterner joined #parrot
14:52 mtk joined #parrot
14:52 pjcj joined #parrot
14:52 jsut joined #parrot
14:52 autark joined #parrot
14:52 brambles joined #parrot
14:52 particle joined #parrot
14:52 atrodo joined #parrot
14:52 masak joined #parrot
14:52 jtpalmer joined #parrot
14:52 GeJ joined #parrot
14:52 pmichaud_ joined #parrot
14:52 sorear joined #parrot
14:52 elmex joined #parrot
14:52 nine joined #parrot
14:52 TonyC joined #parrot
14:52 davidfetter joined #parrot
14:52 alester joined #parrot
14:52 bluescreen joined #parrot
14:52 JimmyZ joined #parrot
14:52 preflex joined #parrot
14:52 alvis joined #parrot
14:52 cotto joined #parrot
14:52 tadzik joined #parrot
14:52 PerlJam joined #parrot
14:52 szbalint joined #parrot
14:52 awwaiid joined #parrot
14:52 cxreg joined #parrot
14:52 Timbus joined #parrot
14:52 knewt joined #parrot
14:52 wendar joined #parrot
14:52 wagle joined #parrot
14:52 Util_ joined #parrot
14:52 dalek joined #parrot
14:52 arnsholt joined #parrot
14:52 NotFound joined #parrot
14:52 Kovensky joined #parrot
14:52 rblackwe joined #parrot
14:52 japhb joined #parrot
14:52 benabik joined #parrot
14:52 whiteknight joined #parrot
14:52 lucian joined #parrot
14:52 Khisanth joined #parrot
14:52 Coke joined #parrot
14:52 mj41 joined #parrot
14:52 jevin joined #parrot
14:52 simcop2387 joined #parrot
14:52 Maddingue joined #parrot
14:59 aloha (parrot/parrot) Issues opened : 763 (api_string_export_wchar allocated one too less) by rurban : https://github.com/parrot/parrot/issues/763
15:05 dalek parrot: df12f34 | rurban++ | src/embed/strings.c:
15:05 dalek parrot: api_string_export_wchar allocated one too less
15:05 dalek parrot:
15:05 dalek parrot: clang -faddress-sanitizer detected this heap-buffer-overflow in
15:05 dalek parrot: t/src/embed/strings.t, allocated 173 bytes only, writing to cstr[173].
15:05 dalek parrot: review: https://github.com/parrot/parrot/commit/df12f34e94
15:05 dalek parrot: 8bd397a | rurban++ | include/parrot/compiler.h:
15:05 dalek parrot: avoid PARROT_NO_ADDRESS_SAFETY_ANALYSIS redefinition warnings
15:05 dalek parrot: review: https://github.com/parrot/parrot/commit/8bd397ab95
15:05 dalek parrot: ca27ffa | petdance++ | / (2 files):
15:05 dalek parrot: Merge pull request #763 from rurban/master
15:05 dalek parrot:
15:05 dalek parrot: api_string_export_wchar + PARROT_NO_ADDRESS_SAFETY_ANALYSIS redef
15:05 dalek parrot: review: https://github.com/parrot/parrot/commit/ca27ffa452
15:05 alester rurban++
15:06 dmalcolm joined #parrot
15:07 rurban one more is coming from make test. t/pmc/structview.t test 5
15:08 rurban Parrot_StructView_set_integer_keyed_orig
15:08 moritz rurban: are you hunting address violations?
15:09 aloha (parrot/parrot) Issues closed : 763 (api_string_export_wchar allocated one too less) by rurban : https://github.com/parrot/parrot/issues/763
15:12 rurban yes, just asan runs
15:14 schmooster joined #parrot
15:15 whiteknight rurban++
15:15 dalek parrot: d3dcc4b | petdance++ | src/embed/strings.c:
15:15 dalek parrot: localized and consted some local variables. Parenthesized another malloc call like in df12f34e9465531a89853ce541096fefc7025e0f
15:15 dalek parrot: review: https://github.com/parrot/parrot/commit/d3dcc4b14f
15:16 alester What is "asan"?
15:16 alester Oh, oh, got it.
15:16 alester I can't get a llvm/clang build working on my box.
15:18 rurban alester: what's not in the build instructions there: remove a prev. system clang first.
15:18 alester I don't think I have one.
15:19 alester You're on Mac OS X, right?  I've been doing it on Linux
15:19 rurban an old clang can never boootstrap a new one
15:19 rurban I'm doing both
15:19 alester Huh.
15:19 rurban now linux
15:19 estrabd joined #parrot
15:19 alester oh well, i have plenty of other projects I'm working on, and it's even less important now that you're on the case.
15:19 rurban mac osx is trickier you need no-pie
15:19 alester "no-pie"?
15:19 rurban like PIC
15:20 alester I don't understand.
15:27 plobsing joined #parrot
15:27 rurban no, better like address randomization. http://developer.apple.com/library/mac/#docu​mentation/Darwin/Reference/Manpages/man1/ld.1.html
15:28 rurban similar to windows ASLR
15:35 alester rurban: I'm surprised your talk got rejected for YAPC::NA.  Are you coming anyway?
15:37 rurban sure, I'm coming. My company pays for it :)
15:38 alester I'm going to drop JT a line saying "If anyhting opens up, please run rurban's talk"
15:39 whiteknight what is his talk about
15:40 alester https://github.com/parrot/parro​t/pull/763#issuecomment-5513148
15:40 Coke rurban++ # my boss rejected my request. Of course, we don't use perl at work, so I can't complain. ;)
15:40 rurban http://code.google.com/p/addr​ess-sanitizer/wiki/FoundBugs see perl
15:40 alester I'm just taking days off.  Madison is only 2 hours from my house.
15:41 rurban and how it works overall, and how to analzye the found bugs. clang is not easy to debug with gdb. most vars are in registers
15:41 rurban and that you can just use it and forget about it, since it's so fast.
15:41 alester rurban, what TZ are you?  I'm guessing UTC+0100?
15:42 alester I'd love to work on this more w/you but I don't have the time right now.
15:42 rurban I'm in Texas/Houston
15:42 alester oh, huh, why'd I think you were in .fr?
15:43 rurban I moved over, working for cPanel at the perl5 compiler
15:43 alester Aha.  You're a cpaneler.  Now it makes sense.
15:43 rurban we use the compiler all day long,
15:44 alester cpanel: The company with the biggest logo on the YAPC::NA T-shirt. :-)
15:44 alester I was just at a planning meeting on Tuesday.
15:44 rurban like facebook and google you need to have some technological advantage :)
15:44 mdupont joined #parrot
15:46 alester rurban: Are you in the cpanel recruitment video?
15:46 rurban I also use asan all day long here
15:46 rurban Nope, I don't think so.
15:46 alester rurban: Aha, so you WERE in .fr.  I'm not crazy!
15:46 rurban No, I'm from Austria, same town as Arnold Schwarzenegger
15:47 rurban But I'm a better surfer than him
15:47 alester Austria, France, to us Americans it's all the same.
15:47 rurban We also have french folks here at cPanel
15:48 Psyche^ joined #parrot
15:48 alester I get the impression that you guys are sending pretty much the whole office up to Madison.
15:49 rurban We are about 200 and we are /only/ coming with 40 :)
15:49 rurban I guess Madison must be pretty
15:50 rurban http://www.kiplinger.com/slideshow/more-​great-cities-for-young-adults/9.html#top
15:50 alester Madison is indeed very nice.
15:51 rurban maybe that's why everybody wants to come. I have no idea. It's pretty far away from everywhere
15:51 alester I mean, if you don't mind snow.
15:52 alester and the Green Bay Packers
15:52 rurban Last years Ashville was worst city for singles http://www.kiplinger.com/slidesho​w/worst-cities-for-singles/4.html
15:53 rurban But I liked it nevertheless. It's not that you are looking for one night stands at perl conferences
15:55 rurban unaligned FixedIntegerArray is also broken a bit
15:55 alester rurban: But they seem to happen anyway! :-)
15:55 alester rurban: Check yr mailbox, you're on the cancellation list for YAPC
16:02 rurban No mail yet.
16:03 alester rurban@x-ray.at ?
16:03 rurban yes
16:03 alester spam filter?
16:03 alester It's mostly quoted text
16:03 rurban git it, thanks :)
16:03 rurban got it
16:04 rurban I think my talk scared them being too technical
16:14 rurban tempc in structview.pmc:613 is off by one
16:14 rurban Maybe that's why the hi-bit is sometimes wrong (int overflow errors)
16:20 jashwanth joined #parrot
16:43 cotto ~~
16:43 dalek rakudo/nom: 637197e | moritz++ | t/spectest.data:
16:43 dalek rakudo/nom: run new test file
16:43 dalek rakudo/nom: review: https://github.com/rakudo/rakudo/commit/637197eacb
16:46 dalek rakudo/nom: ef715f0 | moritz++ | t/spectest.data:
16:46 dalek rakudo/nom: [t/spectest.data] bring back in asciibetical order
16:46 dalek rakudo/nom:
16:46 dalek rakudo/nom: also delete reference to outdated S07 test
16:46 dalek rakudo/nom: review: https://github.com/rakudo/rakudo/commit/ef715f03ce
16:48 PacoAir joined #parrot
16:48 rurban How about #if 0 in parrot code? for the sake of understanding
16:48 whiteknight what do you mean?
16:49 rurban unaligned access in StructView_set_integer_keyed is off-by-one, but you would not understand it without printing out the access patterns
16:50 rurban At least I do. But I can also write a comment.
16:50 rurban I send you a pull request and you can decide. At least I fixed the bug.
17:03 dalek parrot: 1e5f20e | rurban++ | src/pmc/structview.pmc:
17:03 dalek parrot: avoid invalid unaligned StructView_set_integer_keyed access
17:03 dalek parrot:
17:03 dalek parrot: clang -faddress-sanitizer found this invalid read access in two scenarios:
17:03 dalek parrot: 1: cptr[8/8] with uint1
17:03 dalek parrot: 2: off-by-one access on multi-byte values
17:03 dalek parrot: review: https://github.com/parrot/parrot/commit/1e5f20eefc
17:03 dalek parrot: a850da6 | petdance++ | src/pmc/structview.pmc:
17:03 dalek parrot: Merge pull request #764 from rurban/master
17:03 dalek parrot:
17:03 dalek parrot: last found asan bug (I hope)
17:03 dalek parrot: review: https://github.com/parrot/parrot/commit/a850da6652
17:06 rurban Now let's see if parrot with perl6 survives this treatment :)
17:14 moritz is -faddress-sanitizer now enabled by default on clang builds?
17:14 moritz or is that something you add to the compiler options?
17:15 rurban debian unstable I think so
17:15 rurban It's in llvm 3.1 mainline since about a month or so.
17:15 rurban I usually compile from source
17:16 rurban C is pretty stable, there are some bugs in the Windows port and C++
17:16 rurban bugs = false positives
17:17 rurban One more in parrot: heap-use-after-free in t/run/options.t
17:22 rurban 34 --leak-test fails
17:23 contingencyplan joined #parrot
17:28 dukeleto ~~
17:31 rurban Oh my, Parrot_x_exit needs PARROT_NO_ADDRESS_SAFETY_ANALYSIS added
17:33 rurban nope, did not help
17:34 rurban But it only fails with --leak-test, so it's probably not important
17:35 whiteknight That function is having problems?
17:35 whiteknight It should just be some bookkeeping and a longjmp
17:36 rurban With --leak-test at the final interp->exit_handler_list = NULL; interp is already freed
17:36 whiteknight oh wow
17:36 whiteknight do you have a backtrace for that?
17:36 rurban But only with --leak-test. Maybe we should check for this flag and do not write then
17:37 whiteknight we probably need to rewrite the logic for that flag
17:37 nopaste "rurban" at 208.74.121.102 pasted "With --leak-test at the final interp->exit_handler_list = NULL; interp is already freed" (13 lines) at http://nopaste.snit.ch/137744
17:39 nopaste "rurban" at 208.74.121.102 pasted "asan warning" (37 lines) at http://nopaste.snit.ch/137745
17:39 whiteknight Hmm... that definitely doesn't look right
17:40 whiteknight I think what we want there is to call the at_exit handlers and just return 0
17:40 rurban calling Parrot_x_jump_out is also not kosher
17:40 whiteknight no, we don't want to do either there
17:45 aloha (parrot/parrot) Issues opened : 765 (With --leak-test at the final interp->exit_handler_list = NULL; interp is already freed ) by rurban : https://github.com/parrot/parrot/issues/765
17:56 whiteknight rurban: If you're not going to fix that ticket, assign it to me. I can do it tonight
17:57 whiteknight if you want it, that's fine too :)
17:57 rurban I would if I can. Could not find how to assign
17:58 dalek rakudo/nom: bc520af | moritz++ | t/spectest.data:
17:58 dalek rakudo/nom: run caller.t test
17:58 dalek rakudo/nom: review: https://github.com/rakudo/rakudo/commit/bc520af168
17:59 rurban I'd rather not check the fragile gc logic. Yesterday was too frightening already.
17:59 whiteknight I'll assign the ticket
18:00 whiteknight I'm going to need to learn how to use this analyzer so I can properly test it
18:00 whiteknight rurban: If you can't give a talk at YAPC, could you write up a short How-To about it for parrot-dev?
18:01 rurban cat .gdbinit
18:01 rurban break __asan_report_error
18:01 rurban That's the best hint
18:01 whiteknight ok
18:01 rurban Sure. I'll add to my blog post series about asan  http://blogs.perl.org/users/rurban/20​12/03/address-sanitizer-round-2.html
18:03 whiteknight http://nostarch.com/newsletters/dayagainstdrm.htm
18:04 rurban Whow, super offer! thanks
18:05 whiteknight excellent blog post
18:05 whiteknight (I saw that link in my RSS feed and figured I would pass it along to parrot devs)
18:06 rurban I didn't yet wrote about the CPAN findings (DBI and more), and parrot.
18:07 whiteknight okay, I look forward to it
18:13 rurban If this guy would have used asan he would have been much faster coming up with his exploits: http://nostarch.com/download/bughunter_ch2.pdf
18:20 dalek rakudo/nom: 8fff4e7 | moritz++ | t/spectest.data:
18:20 dalek rakudo/nom: run counted.t test file
18:20 dalek rakudo/nom: review: https://github.com/rakudo/rakudo/commit/8fff4e7d65
18:24 mtk joined #parrot
18:27 plobsing joined #parrot
18:34 mtk joined #parrot
18:39 fperrad_ joined #parrot
18:50 brrt joined #parrot
18:51 rurban make install: -rw------- 1 root 33576 May  4 13:27 /usr/local/lib/parrot/4.3.0-devel/parrot_config.o
18:51 rurban /usr/bin/ld: error: cannot open /usr/local/lib/parrot/4.3.0-devel/parrot_config.o: Permission denied
18:57 mtk joined #parrot
18:57 aloha joined #parrot
18:57 perlite joined #parrot
18:57 bacek_at_work joined #parrot
18:57 dukeleto joined #parrot
18:57 dngor joined #parrot
18:57 eternaleye joined #parrot
18:57 sri joined #parrot
18:57 ttbot joined #parrot
18:57 jlaire joined #parrot
18:57 Tene joined #parrot
18:57 Infinoid joined #parrot
18:58 Infinoid joined #parrot
18:58 Tene joined #parrot
18:58 jlaire joined #parrot
18:58 ttbot joined #parrot
18:58 sri joined #parrot
18:58 eternaleye joined #parrot
18:58 dngor joined #parrot
18:58 dukeleto joined #parrot
18:58 bacek_at_work joined #parrot
18:58 perlite joined #parrot
18:58 aloha joined #parrot
18:58 mtk joined #parrot
18:58 Patterner joined #parrot
18:58 dmalcolm joined #parrot
18:58 pjcj joined #parrot
18:58 jsut joined #parrot
18:58 autark joined #parrot
18:58 brambles joined #parrot
18:58 particle joined #parrot
18:58 atrodo joined #parrot
18:58 masak joined #parrot
18:58 jtpalmer joined #parrot
18:58 GeJ joined #parrot
18:58 pmichaud_ joined #parrot
18:58 sorear joined #parrot
18:58 elmex joined #parrot
18:58 nine joined #parrot
18:58 TonyC joined #parrot
18:59 brrt joined #parrot
18:59 plobsing joined #parrot
18:59 contingencyplan joined #parrot
18:59 mdupont joined #parrot
18:59 benabik joined #parrot
18:59 whiteknight joined #parrot
18:59 Khisanth joined #parrot
18:59 Coke joined #parrot
18:59 mj41 joined #parrot
18:59 jevin joined #parrot
18:59 simcop2387 joined #parrot
18:59 Maddingue joined #parrot
19:00 davidfetter joined #parrot
19:00 alester joined #parrot
19:00 bluescreen joined #parrot
19:00 preflex joined #parrot
19:00 alvis joined #parrot
19:00 cotto joined #parrot
19:00 tadzik joined #parrot
19:00 PerlJam joined #parrot
19:00 szbalint joined #parrot
19:00 awwaiid joined #parrot
19:00 cxreg joined #parrot
19:00 Timbus joined #parrot
19:00 knewt joined #parrot
19:00 wendar joined #parrot
19:00 wagle joined #parrot
19:00 Util_ joined #parrot
19:00 dalek joined #parrot
19:00 arnsholt joined #parrot
19:00 NotFound joined #parrot
19:00 Kovensky joined #parrot
19:00 rblackwe joined #parrot
19:00 japhb joined #parrot
19:04 dalek rakudo/nom: cf40f24 | moritz++ | src/core/Str.pm:
19:04 dalek rakudo/nom: [Str.match] add :th as alias for :nth
19:04 dalek rakudo/nom: review: https://github.com/rakudo/rakudo/commit/cf40f24d09
19:22 lucian joined #parrot
19:23 bluescreen joined #parrot
19:59 dalek parrot: 6c302cf | jkeenan++ | src/pmc/structview.pmc:
19:59 dalek parrot: [codingstd] Eliminate hard tabs and make indentation nice.
19:59 dalek parrot: review: https://github.com/parrot/parrot/commit/6c302cf6ac
20:16 rich joined #parrot
20:16 rich left #parrot
20:22 travis-ci joined #parrot
20:22 travis-ci [travis-ci] parrot/parrot#290 (master - 6c302cf : jkeenan): The build was broken.
20:22 travis-ci [travis-ci] Change view : https://github.com/parrot/par​rot/compare/a850da6...6c302cf
20:22 travis-ci [travis-ci] Build details : http://travis-ci.org/parrot/parrot/builds/1248054
20:22 travis-ci left #parrot
20:37 dalek nqp/toqast: 6f91998 | jnthn++ | src/QAST/Block.nqp:
20:37 dalek nqp/toqast: Generate compilation-unit unique IDs for QAST::Blocks (or allow explicit setting of one).
20:37 dalek nqp/toqast: review: https://github.com/perl6/nqp/commit/6f91998d96
20:37 dalek nqp/toqast: feb325c | jnthn++ | src/QAST/Compiler.nqp:
20:37 dalek nqp/toqast: Code-gen for outer chain.
20:37 dalek nqp/toqast: review: https://github.com/perl6/nqp/commit/feb325c91c
20:37 dalek nqp/toqast: 5f34bad | jnthn++ | src/QAST/Compiler.nqp:
20:37 dalek nqp/toqast: Generate with :anon flag to avoid a Parrot namespace installation.
20:37 dalek nqp/toqast: review: https://github.com/perl6/nqp/commit/5f34bade11
21:00 dalek nqp/toqast: 7de618b | jnthn++ | src/QAST/ (2 files):
21:00 dalek nqp/toqast: First crack at code-gen for :op('call').
21:00 dalek nqp/toqast: review: https://github.com/perl6/nqp/commit/7de618b82f
21:25 benabik ~~
21:27 plobsing joined #parrot
21:28 cotto ~,
21:50 rurban thanks jkeenan. I just installed the .el files for parrot for proper formatting
22:09 bluescreen joined #parrot
22:41 davidfetter joined #parrot
23:05 whiteknight joined #parrot
23:16 whiteknight good evening, #parrot
23:30 dalek Heuristic branch merge: pushed 580 commits to parrot/mem_limit by Whiteknight
23:31 whiteknight benabik: ping
23:31 benabik whiteknight: pong
23:31 whiteknight benabik: the nqp_pct branch, what's the status of it?
23:32 benabik whiteknight: Does PIR very slowly, does very little bytecode.
23:32 whiteknight benabik: okay. Is it something we're going to pursue still, or is it completely supplanted by pact?
23:33 benabik whiteknight: It should probably be replaced with QAST and PACt.
23:34 whiteknight ok
23:34 whiteknight Otherwise I would try to bring it up to date with master or something
23:34 benabik The speed was so painful.
23:34 whiteknight ok
23:40 dalek parrot/manual_args: de203dd | Whiteknight++ | / (8 files):
23:40 dalek parrot/manual_args: Add in a :manual_args flag to PIR, for testing
23:40 dalek parrot/manual_args: review: https://github.com/parrot/parrot/commit/de203dde41
23:40 dalek parrot/manual_args: d7856f8 | Whiteknight++ | / (217 files):
23:40 dalek parrot/manual_args: Merge from master. Fix a conflict in IMCC with handling of automagic 'self'. Add some comments too, to explain the code after the merge
23:40 dalek parrot/manual_args: review: https://github.com/parrot/parrot/commit/d7856f8b7f
23:46 whiteknight I'm trying to dig through some of my old branches and either clean them or update them
23:46 whiteknight I have far too many old branch refs on my machine, idle
23:49 benabik Yeah.
23:50 benabik Oh.  Forgot pct_winxed.  IMHO, that's a better direction for PCT to head.
23:56 whiteknight ok

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

Parrot | source cross referenced