www.parrot.org/ | 283 RTs left | Next release: 2009-07-21
Set by moderator on 19 July 2009.
pmichaud Good evening, #parrot (from OSCON) 00:01
jhorwitz ~~ 00:07
00:12 Whiteknight joined
Whiteknight kid51++ # All his ticket work! 00:13
00:20 patspam joined
dalek TT #856 created by jkeenan++: segfault in pbc_merge 00:20
chromatic pong 00:25
dalek rrot: r40162 | whiteknight++ | trunk/src (2 files):
[oo] Add some documentation about C3 and MRO as per kid51++'s suggestion for RT #46101.
jhorwitz chromatic: my 1.4 milestone calls for listing the parrot API functions that will be decorated with PARROT_API. do we want to just list them now or actually tack on PARROT_API and make the appropriate headerizer changes? 00:27
chromatic Tuesday's near. cotto makes the call, I think. 00:28
Updating the PDD should be sensible enough, if all of these functions 1) exist and 2) have PARROT_EXPORT anyway. 00:29
jhorwitz yeah -- i've also made local changes for PARROT_API just to test, but i can break that off into a branch for the next release.
bacek_at_work good morning #parrot 00:45
dalek TT #857 created by jkeenan++: Parrot_floatval_time() and Parrot_intval_time() do not match up on Win32 00:47
00:51 dukeleto joined 00:52 dukeleto joined 00:58 Casan joined
Casan a look at the parrot roadmap on trac shows that tickets keep getting pushed from milestone to milestone, and most of the 18 missing for 1.4 will likely be pushed to 1.5 which will then have 39 tickets.. I'm just wondering, is there a special issue which pushing dev to a halt at the moment, seems like after 1.0 and the 1.0 hague grant mission, things have slowed down. am I missing something? 01:02
moderator www.parrot.org/ | 279 RTs left | Next release: 2009-07-21 01:04
kid51 Casan: I disagree with your characterization of Parrot development as having been 'pushed to a halt at the moment'. 01:07
Yes, some issues do get pushed to a future milestone. (a) Because some of the plans we made last year were unrealistic. (b) Because solving some of these problems is, ummm, hard. 01:08
chromatic (c) Because estimating volunteer work is very difficult. 01:09
kid51 Casan: But issues that are considered to be 'blockers' are discussed in our weekly online #parrotsketch meetings.
Casan kid51: to be honest, I myself have been caught up elsewhere, and have only followed dev from the side, and that line was probably also more harsh than intended, but wearing the external hat for once that would not be a far fetched point of view.
kid51 Casan: Are there particular areas in which you might contribute?
... or with which you are particularly concerned?
01:10 mokurai joined
kid51 One of the things we try to do (perhaps with not great success) is to match: (Volunteer time; Volunteer skill sets) => (Tasks needing attention). 01:11
Casan kid51: mine would be marketing, coordination, QA, as a programmer I'm just a user. I do understand the issues involved.
kid51 Can you do any testing? Do you have any access to OSes, platforms, etc., that are not well represented in our Smolder reports? 01:12
Casan chromatic: fully understandable. been there. and don't get me wrong, I'm definately not bashing, just curious.
kid51: only freebsd and win32 01:13
kid51 What C-compilers do you use on win32?
chromatic A lot of tasks are blocking on a long-delayed refactoring.
Casan kid51: but I could also turn it around, by asking what don't we have covered already, and then build the reasources to head that way. 01:14
kid51 We can always use a steady stream of Win32 test results done with a given compiler: MSVC; gcc on Windows; mingw; etc.
chromatic: Which "long-delayed refactoring" are you referring to? 01:15
chromatic PCC rewiring.
kid51 PCC being? 01:16
chromatic Parrot Calling Conventions
purl Parrot Calling Conventions is probably my best bet. But I don't know if anyone's done tests yet.
kid51 ah, yes.
kid51 brain gets confused with all these abbreviations beginning with 'P'
01:18 Andy joined
kid51 Casan: If you are capable of studying a C-function and writing one sentence of documentation describing what it does, we have *hundreds* of such functions needing documentation. 01:19
Casan kid51: sounds like a good place to start. 01:20
kid51 Casan: I should also note that summing up the total of tickets with 1.4 and 1.5 milestones produces an artificially large result (more)
... because there are a number of tickets which represent the final elimination of deprecated items that can only be done once 1.4 -- a six-month deprecation point -- is past. 01:21
So, e.g., I have 5 tickets, none of which I can commit code for and close until Tuesday -- but after then all 5 will go poof.
01:21 theory joined
Casan kid51: as for your other suggestion with the win32 tests, and the code review/documentation part, then I was just looking at the "how to get involved" list, where I actually expected a "what we need to evolve faster, better" with the suggestion you just gave.. but didn't find them there.. are they written elsewhere? 01:22
kid51 Casan: valid point; I don't think they are elsewhere 01:23
re C function documentation: Run 'prove -v t/codingstd/c_function_docs.t' to see what needs documentation.
"what we need to evolve faster, better" is 10 more advanced C programmers willing to donate 10-20 hours weekly to Parrot ;-) 01:24
Casan kid51: my relation so far is that I have read about anything I could find for years and years on parrot, and perl, maintain an in-house developed cms in p5, and I'm only a basic programmer, though I know the concepts, but I spend most of my time in management and marketing, so I'm not the sharpest programming tool, but one bite at a time I can do.. 01:25
Whiteknight chromatic: is there any new news about pcc_rewiring? I would like to bump that up to high priority after 1.4 is released 01:26
kid51 Casan: Well, my own fu beyond P5 is limited ... but take a look at our Smolder reports and think about submitting. 01:27
chromatic It's allison's.
In theory, she'll have time after OSCON.
kid51 purl: attending OSCON?
purl kid51: wish i knew
chromatic I'm not planning anything until it lands. 01:28
kid51 purl: OSCON
purl OSCON is, like, july 23 to 27 in portland, or at conferences.oreillynet.com/os2007/ or July 21-25, 2008 in Portland, OR or www.conferences.oreilly.com/oscon/ or rumored to move to san jose next year :(
Tene kid51: thank you for your ticket work. i've commented on the tickets relevant to me (exceptions).
kid51 Tene you're welcome
Whiteknight when is OSCON?
chromatic Starts tomorrow.
Whiteknight oh, so soon
Tene Whiteknight: allison told me that it was very high priority for her shortly after 1.4
Whiteknight okay, that's exactly what I wanted to hear 01:29
Tene Me too.
Casan kid51: ok so if we could do some more efforts in getting more advanced C programmers, that should help. thats another item for the "what we need list" now thats something to relate to. ok, will save that, and think about what to do in regards to these suggestions..
Whiteknight because I have a lot of plans that are blocking on that, although I also have a few projects that are unrelated too
kid51 purl OSCON 01:30
purl hmmm... OSCON is the O'Reilly Open Source Convention, July 20-24, 2009, in San Jose, CA, USA (en.oreilly.com/oscon2009)
kid51 Casan: Thanks. 01:31
Whiteknight goodnight
kid51 purl attending OSCON
purl attending OSCON is pmichaud
kid51 purl attending OSCON 01:36
purl i heard attending OSCON was pmichaud or Andy
Andy ME! ME! 01:39
kid51 Andy r u there yet? 01:40
Andy nope
kid51 When do us poor folks get to hear "Just Enough C For Open Source Projects"?
Andy fly me out to NY
chromatic Hm, that might make a nice book. 01:43
Casan reading through the slides from oscon2008 already. a nice pointer for me, good timing. 01:44
Andy chrmoatic I was thinking that 01:45
Casan: Tuesday I do it as a 3-hour tutorial
Casan Andy: Tuesday I'm in Denmark, busy working, preparing for moving to France to study/work. 01:46
Andy: any chance the 3 hour tutorial will be recorded? 01:47
Andy none
O'reilly charges big bucks for tutotials. They're not going to record + give away. 01:49
Casan Andy: its alright. 01:50
actually that brings up a thought (although my intention is not to challenge the business model), but one of the reasons ruby got sexy was the free video tutorials en mass. does the pro training biz in perl have a natural dilemma with this? 01:52
dalek TT #858 created by jkeenan++: HLL exception handling 01:54
moderator www.parrot.org/ | 277 RTs left | Next release: 2009-07-21 01:55
Infinoid kid51++ 01:55
02:03 theory joined
dalek TT #825 closed by jkeenan++: PATCH: Fix incorrect references in pdd22 02:08
rrot: r40163 | jkeenan++ | trunk/docs/pdds/pdd22_io.pod:
Applying patch correcting references to source code files supplied by jrtayloriv++ in trac.parrot.org/parrot/ticket/825.
 02:09
02:11 TiMBuS joined
dalek TT #859 created by jkeenan++: 'make splint' needs graceful failure message 02:32
02:35 janus joined
cotto Heh. Thunderbird flagged the commit message for r40162 as a possible scam. 02:52
chromatic, what's this PARROT_API business I'm supposed to make the call on? 02:55
chromatic If it's worthwhile or safe to edit source files and headers to add annotations to API functions. 02:56
cotto I'm not familiar with this issue. Would this just be to help generate documentation? 02:58
since we already have PARROT_EXPORT
chromatic I'm not sure exactly. I'd like to have a way to enforce things, but we're talking about tagging a lot of functions. 02:59
cotto btw, is it safe to reject any TGE improvement tickets?
chromatic allison knows better there. 03:00
03:00 dukeleto joined
cotto Is OSCON going to interfere #ps this week? 03:01
s/re/re with/
chromatic Only for the attendees.
cotto any idea who's going? 03:02
kid51 purl attending OSCON
purl hmmm... attending OSCON is pmichaud or Andy
kid51 must sleep
purl $kid51->sleep(8 * 3600);
cotto going to OSCON?
purl going to OSCON is rjbs or perigrin's boss and boss's boss or pjf or jarich or nothingmuch and obra or tchrist
cotto OSCON?
purl i think OSCON is the O'Reilly Open Source Convention, July 20-24, 2009, in San Jose, CA, USA (en.oreilly.com/oscon2009)
cotto no worries. It's not critical to close old RT tickets before the releae. 03:03
03:12 mokurai joined
bacek_at_work "Cristoph Otto, Software Test Engineer, Microsoft Open Source Software Lab" 03:54
sounds familiar :)
cotto LinkedIn? 04:12
purl O SPPM tem um grupo de profissionais no LinkedIn: www.linkedin.com/groups?gid=64857 Você já se cadastrou lá?
cotto javascript.crockford.com/www_svendt...index.html - so...many...commas
bacek_at_work, My LinkedIn profile should reflect reality more accurately now. 04:13
bacek_at_work cotto: I did check it :) 04:14
cotto Where'd you see that then?
chromatic Alias's journal 04:15
purl Alias's journal is use.perl.org/~Alias/journal/
bacek_at_work Alia^W What chromatic said 04:16
cotto It's nice to get something like my name out there ;) 04:17
bacek_at_work cotto: Hey! I was mentioned in one of chromatic's posts! Why do you need more? :)
Andy What am I doing wrong here? 04:26
erxz.com/pb/18875
why does no sorting happen?
cotto is sad he won't be able to see "The Conway Channel" at OSCON. 04:28
chromatic strcmp(3) arguments are "pointers to char"
return strcmp(* (char * const *) p1, * (char * const *) p2); 04:29
Apparently you have to write a strcmp wrapper.
Andy yeah, but void should be the same
note that the reverse run reverses the order 04:30
chromatic It should be the same, but gcc warns about it, so it doesn't think it's the same.
Andy is it comparing actual pointer values, not the strings?
chromatic Maybe glibc inlines something.
That could be too. 04:31
Andy yeah, but I'm ignoring that warning
digging.
TonyC strcmp() wants names[i], called like that it receives names+i
where i is the index being compared
Andy somehow it's comparing pointer values, not pointer contents 04:34
chromatic Right, qsort() calls its functions with pointers to pointers to char. 04:36
Andy no, that's not right. At least that's not how I understand it, or read the docs 04:38
qsort(void *base, size_t nmemb, size_t size, int (*compar)(const void *, const void *));
chromatic man 3 bsearch 04:39
Andy yes, and?
purl and is the register frame something like the Sparc's?
Andy what are you seeing there
TonyC if you were sorting an int array the parameters would be int *, you're sorting an array of const char *, so the parameters are const char * * 04:40
chromatic Yep, you need the dereference. 04:41
Andy there we go 04:42
thank you 04:43
TonyC reverse_nocase looks wrong too
Andy yup 04:44
fixed erxz.com/pb/18878 04:46
mmm, consting 04:48
what a fine teaching moment this will provide on Tuesday 04:52
"Do you hate pointers yet? Let's turn to the next slide."
chromatic Don't hate pointers. Hate pass-by-value. 04:54
bacek_at_work [14:54:21] Tom: and then on the next slide 04:55
[14:54:25] Tom: HAHHA YOU FOLLOWED A POINTER!!1
:)
chromatic: pass-by-value is good optimisation strategy
chromatic Until you do type erasure. 05:12
cotto Should we deprecate Iterator instantiation in favor of the iter opcode? 05:17
or does Iterator instantiation dtrt
nm; dtrt 05:18
05:47 theory joined
bacek_at_work cotto: let's deprecate it 05:53
Current Iterator.init_pmc is awful.
cotto +1 06:04
purl 1
cotto it's ugly
06:37 iblechbot joined
cotto I'd like to reiterate the request that all commit messages be prefixed with the effected subsystem or task in square brackets. 06:53
japhb, ping 06:54
msg japhb How would you summarize your OpenGL changes since the last release? 07:03
purl Message for japhb stored.
dalek rrot: r40164 | cotto++ | trunk/NEWS:
[NEWS] likely inadequate NEWS update for the 1.5 release - if you see any holes, fill them!
 07:06
cotto Gah. It's obviously inadequate for 1.5. me-- 07:08
moritz isn't it 1.4? 07:09
cotto Yes. 07:12
That's why.
moritz ah. 07:13
cotto At least I managed to get NEWS right.
moritz nobody reads commit messages anyway
cotto Or maybe I'm FROM THE FUTURE. 07:14
moritz some friends of mine cited in their thesis the well known A. Nonymous, "Nobody reads references anyway"
cotto That's an excellent paper. 07:17
07:18 autark joined
moritz indeed. 07:18
07:24 Zak joined
bacek_at_work cotto: now Hashes can have native keys/values not only string/pmc. 07:25
cotto: Ah. Can we add deprecation notice about nomenclature of VTABLEs?
e.g. get_bignum; set_bignum_*; pow/i_pow; nextkey_keyed*; 07:29
mikehh `BTW - make html still says - Failed to process docs/book/draft/ch03_pir.pod. 07:35
should we correct this or add a dummy draft/ch03 saying that chapter 3 is now book 07:36
cotto bacek_at_work, is that for NEWS? 07:43
and yes, deprecated VTABLE functions should be mentioned in DEPRECATED.pod
bacek_at_work cotto: first for NEWS. Second for DEPRECATION.pod 07:44
cotto +1 then. 07:46
Heh. SSE 4.2 has instructions intended to accelerate XML.
s/XML/XML processing 07:47
moritz SSE? the CPU vectorization thing? 07:49
cotto Yes. 07:51
blog.reverberate.org/2009/07/18/gaz...e-sse-4-2/
It's fairly generic, though.
bacek_at_work, if you're worried about grammar, don't be. I'll do a cleanup pass on NEWS before the release, assuming other people add to it. 08:10
bacek_at_work cotto: I can worry about only one grammar. It usually saved in ".pg" file. 08:13
cotto That's the important one. 08:16
The parser for NEWS is generally quite forgiving.
night 08:18
mikehh codetest fails - All Others PASS (pre/post config, smolder, fulltest) at r40164 - Ubuntu 9.04 amd64 08:20
nopaste "mikehh" at 90.209.69.171 pasted "PATCH for codetest failure at r40164" (30 lines) at nopaste.snit.ch/17329
mikehh codetest PASSes after the patch 08:21
just trailing spaces
bacek_at_work night cotto 08:23
mikehh: I'll commit it later tonight if noone else will do it before. 08:29
chromatic Applied, thanks.
dalek rrot: r40165 | chromatic++ | trunk/lib/Parrot/Pmc2c/PMC/RO.pm:
[lib] Removed trailing whitespace, updated copyright, coalesced one line
 08:31
chromatic It might be easier to give mikehh commit access to fix up little things like this as he finds them. 08:32
(while not precluding him from applying more complex patches in the future)
moritz aye 08:33
mikehh I gotta submit a CLA but my printer died on me - I am getting a new all-in-one today or tomorrow sometime 08:37
rakudo (415514b) on parrot r40164 (spec at r27621) fails t/spec/S06-signature/optional.rakudo - test 11 08:43
moritz I get the same failure 08:44
mikehh spectest runs in 10 minutes - grabs nearly all my processor resources
08:45 masak joined
mikehh 3 procs running at 100% - the other about 85% 08:45
bacek_at_work moritz: any ideas about optional.t failure? 08:46
mikehh used to take over 30 minutes on my dual proc system
moritz bacek_at_work: no 08:47
mikehh the other 3 similar tests are TODOed
moritz bacek_at_work: I did some Rakudo changes over the weekend, but I didn't touch anything that deep
mikehh it passed on r40131 but failed on r40136 08:51
moritz parrot revisions?
mikehh yeah rakudo 3d5f23c PASS - faf91f9 FAIL 08:53
moritz uhm, both 3d5f23c and faf91f9 are doc patches 08:54
mikehh spec (pugs) r27602 PASS r27603 FAIL
I can follow svn revisions - they increment - git??? - I suppose you can look at the history 08:59
moritz ah, it's a test change 09:01
pugs r27603
so it's fine to fudge it
mikehh byw how do you get changes between 3d5f23c... and faf91f9...
s/byw/BTW/
moritz git diff 3d5f23c..faf91f9 09:04
mikehh ah ha - I am still learning git 09:05
moritz ok, fudged the tests in pugs repo now
japhb From NEWS: "substantial optimizations in NCI": What commits were those, and are there numbers somewhere? 09:11
cotto: "Beginnings of OpenGL::Math library; mark OpenGL bindings and library experimental" 09:18
09:32 donaldh joined 09:35 mokurai left
mikehh rakudo (415514b) on parrot r40164 (spec at r27622) make test/make spectest PASS 09:39
10:23 bacek_at_work joined
bacek_at_work seen fperrad 10:36
purl fperrad was last seen on #parrot 16 hours, 41 minutes and 1 seconds ago, saying: In any case, the best way to ship OpenGL is one PBC, not 2. Even when there are 2 PIR files.
10:37 Hinrik joined
Hinrik paste.lisp.org/display/83841 - anyone had this issue? 10:38
right before it died it tried to display the generated Pod instead of writing it to a file, that may be the reason it fails
10:51 MoC joined
Hinrik trac.parrot.org/parrot/ticket/860 10:56
dalek TT #860 created by Hinrik++: Failed Pod extraction halts the build process 10:57
11:07 szabgab joined
dalek TT #861 created by MoC++: mingw32-make fails upon tools\\build\\c2str.pl --all 11:10
mikehh Hinric: do you have perldoc - /usr/bin/perldoc? 11:15
sorry 11:16
Hinrik: do you have perldoc - /usr/bin/perldoc? 11:17
Hinric: also see docs.parrot.org/parrot/latest/html/...s.pod.html 11:19
11:20 donaldh joined
mikehh damn - did it again 11:28
Hinrik: also see docs.parrot.org/parrot/latest/html/...s.pod.html
11:34 payload joined
Hinrik yeah I heave perldoc 11:37
have*
and it's in /usr/bin/ 11:38
as you can see in the output I pasted, those "No documentation for ..." are from perldoc
it builds fine on another machine I have, hm 11:46
11:54 dukeleto joined, whiteknight joined
mikehh Hinrik: from my latest build 12:01
/usr/local/bin/perldoc -ud packfile-c.pod ../src/packfile.c
Perldoc (Pod::Perldoc::ToPod) output saved to packfile-c.pod
Hinrik yeah, it's some issue with my particular machine 12:02
just don't know what... 12:03
mikehh Pod::Perldoc::ToPod maybe?
12:07 bacek joined
bacek o hai 12:08
mikehh hi there
bacek hi mikehh 12:11
whiteknight good morning 12:17
purl Here I am, brain the size of a planet, and all they say is 'Good Morning'
mikehh hi 12:22
purl bonjour, mikehh.
12:26 whoppix joined
dalek rrot: r40166 | bacek++ | branches/orderedhash_revamp:
Branch for deinsanify OrderedHash
 12:30
12:31 tetragon joined 12:38 bobke joined
Hinrik *sigh* the problem seems to have mysteriously disappeared 12:43
Coke msg cotto TT #823 is a release blocker, still 13:05
purl Message for cotto stored.
Coke msg bacek TT #823 is a release blocker, still
purl Message for bacek stored.
13:05 kj joined
bacek Coke: hi. I'm still here. 13:05
Coke msg cotto no other fulltest failures on linux, though.
purl Message for cotto stored.
Coke bacek: the default core seems to be the only one those packfile tests pass on.
bacek Coke: checking. 13:07
Coke: are you sure that you don't have some weird and orphaned foo.pmc in src/pmc? 13:09
Because I can't reproduce this failure with any core... :/
fast/cgoto/cgt/switch 13:10
All passed. Debian/Lenny/i386
Coke bacek: I do have "context.pmc", yes. 13:11
dalek TT #862 created by coke++: remove fulltest_all make target
Coke but it's not built or linked or...
bacek Coke: that's a problem.
Coke ... why?
bacek Configure create different enum_class_Foo
Coke (it's leftover from an svn switch.)
bacek And our PBC are too fragile 13:12
Coke stupid idiotic build systme.
I blame the build system - it shouldn't be adding pmcs that are not in the manifest.
retrying.
bacek Coke: good luck :)
Coke that would explain why those test have been failing for me and no one else for some time. 13:15
s/have/had/
(since they were failing for everyone for a short time.)
bacek Indeed 13:16
dalek TT #863 created by fperrad++: GC Segfault 13:18
bacek OH NOES!
Not GC! Not again!
mj41 Hi. Is there a central server to export 'make cover' html?
tapir2 can run 'make cover' at 40 minutes 13:19
13:20 skids joined
Coke mj41: we have a server for the 13:21
smoke output, but not the cover output, so far as I know.
bacek - have my reply all typed out in the ticket, just waiting for fulltest to finish.
bacek crossing fingers
13:29 xiaoyafeng joined 13:45 PacoLinux joined
NotFound hi 13:49
bacek hi NotFound 13:53
13:59 jhorwitz joined, ruoso joined
bacek Coke: any luck? 14:08
14:10 Z3vil joined 14:15 Z3vil left 14:28 iblechbot joined
dalek TT #864 created by erez++: on Win32 binary install, parrot_config.exe contains false installation ... 14:40
14:43 pancake joined 15:01 Zak joined 15:14 MoC joined
Coke bacek: sorry, dealing with a plumbing crisis. 15:17
msg bacek works fine, yes, thanks.
purl Message for bacek stored.
cotto good morning 15:19
purl And good moroning to you, cotto.
Coke cotto: ignore my prior gloom and doom.
particle is that in response to the plumbing issue being solved? 15:20
15:20 donaldh joined
particle now you're happy, everything's working? 15:20
moritz I think wrt packfile problems 15:21
cotto Coke, you're saying that the packfile issue is ok?
TT #823
dalek TT #823 closed by coke++: packfile pmc tests failing 15:27
NotFound I'm unable to reproduce TT #853 15:28
particle seems we need another ticket for the root cause of coke++'s problems
cotto Agreed, but I'm *really* glad that the root cause is known. 15:29
particle config/gen/makefiles/root.in:655 causing the issue? 15:30
pmichaud Good morning, #parrot 15:31
cotto I'll file a tt on extra .pmc files breaking parrot 15:33
particle hrmm, ok.
lib/Parrot/Pmc2c/Pmc2cMain.pm:95 15:34
unshift @{ $allargsref->{include} },
'.', "$allargsref->{bin}/../..", "$allargsref->{bin}/../../src/pmc", "$allargsref->{bin}/../../src/dynpmc";
pmc2c takes *all* .pmc files in src/pmc/ and compiles them
not just those in MANIFEST
this may be expected behavior 15:35
cotto particle, do you have a quick enough fix that I should bother filing a tt?
particle cotto: no.
the solution to the ticket may be documentation of expected behavior
or it may be a change to existing behavior 15:36
NotFound A warning if some of thew pbc obatined is not present in MANIFEST will be nice. 15:37
particle yeah, it could be because you're creating a new pmc 15:40
which is perfectly valid for developers
15:40 theory joined
particle and a warning seems fine in that context. 15:40
NotFound Yes, a way that let you test and experiment, but don't make easy to forget.
particle makefiles/root.in and pmc2c are in agreement, that src/pmc/*.pmc will be processed 15:41
actually, on platforms with a make engine that supports suffix rules well enough, *.pmc in *any* directory will be compiled 15:43
dalek TT #865 created by cotto++: extra .pmc files in src/pmc cause silent pbc compatibility breaks 15:49
Coke cotto++ 16:09
particle: if you're creating a new PMC, you have to add it to the manfiest. 16:10
so keying off the manifest seems sane.
cotto Given that, I don't think failing on a mismatch is too onerous. 16:13
16:13 payload joined
solarion you guys hear the linux outlaws podcast #102? 16:26
Coke nope. 16:27
tell me more.
solarion Coke: there's a segment about an hour or so in that talks about mono and how parrot needs more love (bkuhn from SFLC)
NotFound I never heard any podcast, and don't have any intention to start now X-)
solarion nothing technical, though 16:28
Coke NotFound: ORLY? I have about 7 days worth of podcasts queued up.
need to aggressively delete some of them, apparently. =-)
solarion disk space is cheap. :) 16:29
I need to do more with parrot
NotFound I read a lot faster that anyone can talk.
s/that/than
solarion I just play them while working
Coke NotFound: nice feature on the latest iphone OS - 2x playback speed on podcasts. =-) 16:30
solarion how does that work?
there are some good music podcasts
NotFound Coke: I have even less intention to buy an iphone than to listen to podcasts :D 16:31
Coke points at notfound and screams like the end of _invasion of the body snatchers_ (the one with leonard nimoy) 16:32
NotFound The only Apple product I had some intention to but was a Newton... and it was the less successful Apple product ever X-)
s/but/buy
Well, maybe the Apple III was worse. 16:33
solarion Was it Bob-unsuccessful?
NotFound Very few people even knows what it was. That must mean something about his success. 16:34
japhb: ping 16:39
Coke NotFound: the iphone is approaching newton usefulness.
modulo the handwriting recognition, of course. 16:40
NotFound Coke: Let's hope it doesn't approach his commercial success ;)
solarion or alchemical explorations 16:42
nopaste "NotFound" at 213.96.228.50 pasted "OpenGL HLL usage test" (92 lines) at nopaste.snit.ch/17332 16:44
NotFound I think the differences between this test and the example static-triangle can explain the problem in TT #853 16:46
Coke msg cotto trac.parrot.org/parrot/report/3 shows 17 tickets for the 1.4.0 release; I suspect at least 15 of those are not getting done. 16:48
purl Message for cotto stored.
whiteknight In general I think random tickets which are assigned to a release don't care much weight
we should probably suggest people not assign new tickets to an upcoming release because it's meaningless 16:49
Coke whiteknight: in reality, the release designations are always meaningless.
whiteknight (unless it's demonstrably a release blocker)
particle notfound: en.wikipedia.org/wiki/File:Newton_e...martha.jpg
whiteknight true
Coke whiteknight: there's no such thing as a release blocker, IME. =-)
it's always 'push to next release'
cotto Coke, yeah. I noticed that earlier.
NotFound particle: good example :D 16:50
Coke whiteknight: 8 of those tickets are /roadmap/ items. None of those are done.
Coke checks to make sure he's not assigned to any of those.
NotFound io.t failures in win32 are easy to fix: just revert the test rewriting in pir 16:52
Coke +1
purl 1
Coke purl, you're a GENIUS!
purl Coke: excuse me?
Coke a genius?
purl a genius is working in CPAN shell
Coke cpan shell?
purl rumour has it cpan shell is just so darned useful. or "perl -MCPAN -e shell;" or TURTLE POWER! or "perldoc CPAN" or the Greatest Thing since Monkeypants or "perl -MCPAN -e 'install Module'" or much more fun if you have a big pipe or needs too much goddamn ram. or (but maybe try CPAN -e 'install CPANPLUS')
Coke CPANPLUS? 16:53
purl CPANPLUS is a generic CPAN build engine or helly as a flake or used for smoking modules or at oss.dwim.org/cpanplus-devel/
Coke flake?
NotFound rumour?
purl rumour has it rumour is that most of those will be in the us
Coke =-)
moritz us?
purl us is still trying to find him.
moritz him?
purl well, him is it!
moritz it?
purl i guess it is difficult, yeah. I was interested in it about 14 or so years ago and then gave up. Sold off most of Kyriel's language references.
NotFound yeah?
16:54 Psyche^ joined
cotto monkeypants? 16:56
purl Oh boy!
Coke hee. 16:57
squee!
17:05 mokurai joined
Coke I can revert the op test, I think. 17:06
dalek rrot: r40167 | NotFound++ | trunk (2 files):
[examples] add a variant of opengl static-triangle with HLL
 17:08
Coke it's not as simple as a reversion. some changes have been made since the test was converted. 17:10
NotFound Shit!
Coke aye.
lesse, one is "works on openbsd", that's easy... 17:11
r39814 is a little harder, but still doable. 17:12
I can try to do it this evening if no one beats me to it.
NotFound Maybe we must have dropped all platform checks, assume it works wherever and check smoke results... but not now, just before the release. 17:13
Coke we could also try a better translation to pir for the test. 17:14
might it be a flushing issue?
(though I think I tried adding a .'flush'() to no avail. 17:15
NotFound I'm almost sure is some problem with stdout, but haven't been able to diagnose it better.
Coke ah. I tried to flush the pipe handle, not stdout. 17:16
17:16 flh joined
NotFound The real problem is that the test is too simplistic, but I lack knowledge about using temp files in tests. 17:17
Coke nope, flushing doesn't help. 17:18
I can probably do the de-conversion and keep fperrad's changes. 17:19
17:22 tomyan joined
Coke msg cotto partcl passes all tests (except for the inferior runloop issues) on 2/3 platforms I have access to. 17:33
purl Message for cotto stored.
Coke (the 3rd is windows, and I have no idea if those tests ever worked there.)
cotto Coke, I'm here. You don't have to msg me. ;) 17:34
Coke ah. likely due to a lack of ICU.
cotto: yes, but if I do it that way I don't have to msg you later!
this way, it's more work for you not me. =-)
cotto I guess I admire your laziness.
Coke ^_ 17:35
er, ^_^
hurm. if I require ICU, I suppose I should squawk if someone tries to build me with a parrot that is not config'd with it. 17:36
Anyone bored and want to write some perl? =-)
cotto "Someone" should fix make html before the release.
Coke what's wrong with it? 17:37
Failed to process docs/book/draft/ch03_pir.pod. ? 17:38
17:38 mj41_ joined
cotto Yeah, that. 17:38
purl Sure, that.
Coke we have MANIFEST-style information in the lib for that. very evil. 17:39
NotFound And failed silently.
Coke fixing shortly.
cotto I don't know if it's because of PseudoPOD or actual invalid POD.
Great. Coke++
Coke cotto: fixed in 40168
the file was removed. 17:40
dalek rrot: r40168 | coke++ | trunk/lib/Parrot/Docs/Section/Parrot.pm:
[cage] the POD file referenced here was removed from the repository.
cotto Shouldn't the other files be renumbered and renamed?
Coke they're in draft. =-) 17:41
I think that means 'not worth the effort.'
cotto Great. I'll call that solved then.
Coke was there a ticket? 17:44
NotFound $ cd docs/html/ ; $ find . -size 0 ; # Uhhhhh 17:45
cotto Heh. I just noticed that, NotFound.
Coke #765. 17:46
NotFound I think there is a ticket about this.
Coke #765.
NotFound Yeah
cotto That'd also be nice to fix before the release.
solarion where's the rakudo irc channel?
just says "Ask on IRC how best to help" 17:47
NotFound Will be nice if make html showed some warning about empty result files.
PerlJam solarion: freenode #perl6
solarion ah, that 'splains it, thanks
Coke oddly, metacommitter_guide doesn't show in lib/ (where 'make html' figures out what to generate) 17:48
yah, I think we just need to add those files to the list of things getting generated. 17:50
also, "lib/Parrot/Docs/Section/Parrot.pm" has a lot of commented out code (use's)
17:52 dukeleto joined
Coke dukeleto: hio 17:52
dukeleto hola 17:56
i sent my parrot CLA in, how do I get a commit bit? 17:57
Coke dukeleto: you need 2 people to say "yes, please." 17:58
have you created a trac account?
dukeleto yes, i have a trac account 17:59
Coke k.
NotFound And you must take into account that mail is a bit slow than email ;) 18:02
Coke he faxed it in, which auto-converts to email.
18:02 dukeleto_ joined
Coke aigh, he's multiplying. 18:02
NotFound Faxed? Last time I read a real paper was required.
Coke nope.
that may predate the foundation. 18:03
NotFound Ah, nice.
pmichaud cla?
purl rumour has it cla is Contributor License Agreement or www.perlfoundation.org/contributor_..._agreement or www.parrot.org/foundation/legal or www.parrot.org/files/parrot_cla.pdf
Coke dukeleto: I'm happy to be your sponsor if you don't have one already. give me a moment to enable commit bits. 18:04
(if you break trunk just before the supported release, I will be most displeased. =-)
dukeleto_ Coke: will definitely not break trunk before a release : ) i will wait until the next release is out the door first :)
cotto Yay! Fresh blood! 18:05
dukeleto_ mmmmmm, fresh blood
Coke verify your trac id for me?
dukel3to Coke: dukeleto 18:06
purl dukeleto is, like, still feeling the effects
NotFound We'll tell the Harkkonen if you do that.
Coke dukel3to: done.
dukel3to shweet
wheels within wheels... 18:07
purl Wheels within wheels in a spiral array. A pattern so grand and complex. Time after time we lose sight of the way. Our causes can't see their effects.
dukel3to purl: nice
purl I'm smooth like that
dukel3to Coke: so now I can commit via svn with my trac id and password? 18:08
Coke that should work. I'm sure you can find a typo in the docs somewhere to fix.
dukel3to and should I be adding myself to CREDITS ? 18:12
18:14 sekimura joined
cotto That's a good idea. 18:16
dukel3to cool 18:17
nopaste "rg" at 93.104.50.113 pasted "PLATFORM updates for the platforms I'm running smoketests on" (24 lines) at nopaste.snit.ch/17336
rg only changed the date. since they are green ;) 18:18
18:21 clinton joined 18:22 mokurai joined 18:25 iblechbot joined 18:30 chromatic joined 18:31 darbelo joined
cotto Hi darbelo 18:33
dalek rrot: r40169 | cotto++ | trunk/PLATFORMS:
[PLATFORMS] platform status updates courtesy of rg++
 18:36
darbelo hi cotto
treed Is the double free infinite loop a known/being worked on bug?
chromatic Yes. 18:37
treed k
treed runs into it all the time.
Multiple Dispatch: No suitable candidate found for 'cmp', with signature 'PP->I' 18:38
Is there some way I can insist that my cmp sub does in fact have that signature?
chromatic What's your cmp's multi declaration? 18:40
treed .sub '_cmp' :method :vtable('cmp') :multi(P,"CardinalArray") 18:41
Initially it didn't have one, I added it trying to get past that.
wait.
Did I recompile after that?
>_>
treed tries to be sure.
Okay, yeah, still fails. 18:42
chromatic What happens if you change P to PMC? 18:44
Coke P should be _ 18:45
treed No change.
treed tries that.
ISTR trying _ first, though.
Really wish Cardinal's parser were faster. 18:46
Coke treed: is it slower than partcl's? =-)
treed Yeah, no change.
Coke: Haven't tried it.
But it takes a good 10-15 seconds to compile a 20 line file. 18:47
Maybe more.
It takes longer to parse Test.rb than it does to completely rebuild cardinal.
5 minutes to run ~80 test files
Tene says he has some ideas about how to fix the parser, but he hasn't written them down for me yet. 18:48
Tene Coke: Yes, cardinal's parser is slower than partcl's 18:49
I have a few examples that take several minutes to parse.
Coke it is quite possible to write slow .pg 18:50
cotto whiteknight, is the win64 build working? 18:51
18:51 athomason joined
Coke treed - this could be related to a bug I opened for partcl. 18:53
whiteknight cotto: sort of
it builds, but doesn't pass all tests
I will open a few tickets eventualy for it
treed Coke: Oh? 18:54
dalek rrot: r40170 | cotto++ | trunk/NEWS:
[NEWS] more news updates for 1.4
Coke treed: trac.parrot.org/parrot/ticket/784 18:55
cotto whiteknight, does that mean an unmodified checkout from svn builds (even with test failures)?
Coke (if so, get a PLATFORMS entry update.) 18:56
whiteknight cotto: I think it does build, yes. I need to check and update it tonight
treed Coke: Except that in my case, I'm explicitly declaring a new op sub. May still be related. 18:57
Maybe indicates that subclasses aren't being searched for that sort of thing? 18:58
cotto Cool. Can you update the platforms entry?
whiteknight cott: I will update it, yes
cotto much obliged
Coke make examples_tests has a segfault. whee. 19:00
treed Coke: Regarding your bug, I wonder what Integer's sub for that is declared as? 19:02
Coke treed: it's in src/pmc/integer.pmc 19:03
(it's a C function, not a PIR sub.)
treed Ah. 19:04
I was wondering if maybe the sig was specifically referencing Integer, but it doesn't seem to be.
dalek rrot: r40171 | coke++ | trunk/PLATFORMS:
Update darwin status.
 19:05
19:11 hydrapheetz joined
treed Has anyone examined that bug in the context of making a new un-subclassed class that implements a vtable op? 19:13
And seeing if that works?
Coke it would be nice if someone who wasn't me updated en.wikipedia.org/wiki/Parrot_Foundation to reflect the positions mentioned here: www.parrotblog.org/2008/06/announci...ation.html 19:14
treed; I have not, no; typically if someone's done something for a ticket, they'd leave a note on the ticket.
treed k
jhorwitz cotto: were you planning on tagging before or after #ps tomorrow? 19:32
whiteknight Coke: what else from that blog post needs to be added to the WP page? 19:35
japhb NotFound: pong. I haven't had a chance to backlog, but I noticed your ping. What's up? 19:39
NotFound japhb: I'm looking at TT #853 19:40
japhb: Have you played with the lua failure, or just commenting in general? 19:41
japhb NotFound: I've mostly been responding to fperrad's comments, and making sure there was as much historical info in the ticket as I knew. 19:42
I have not had a chance to recreate the bug myself.
Damn $paying_job ....
NotFound The problem is that I don't think that reverting the change is a good idea, and even less if I can't chekch that doing that solve the problem. 19:44
I commited an example that shows that using the opengl lib from inside .HLL works. 19:45
japhb Ah, from your last comment on TT #853, it looks like the problem is that I broke the mold -- I use .include for its direct functional purpose, rather than just as a way to read "header" files.
Thus subverting the dichotomy between library/ and include/
NotFound japhb: yes, but my main point is that the load_bytecode form can work, and in fact it does. 19:46
japhb NotFound: fair enough.
I'm looking at your example now.
NotFound If there is some problem with lua, lua must be fixed, not parrot worsened. 19:47
Coke whiteknight: I was thinking the list of board positions and their occupants.
rather than "allison is in charge." =-)
whiteknight Coke: done. 19:50
NotFound japhb: for the berak mold part don't worry, it was common in the repo before we started fixing it.
japhb NotFound: why do you need to explicitly grab the namespace in static-triangle-hll.pir and do "import_gl($P0)"? Why doesn't the defaulting code in OpenGL.pir's _export_all_functions Just Work?
NotFound japhb: uh, I was not sure if it was required, introduced it at the start and forgot to test with it removed. 19:51
Coke whiteknight: Danke. 19:52
NotFound japhb: but I think is clear to explictly show where it is supposed to be exporting to.
japhb If it works for you without, can you pull that out? I'd like minimal differences between the variants, since the static-triangle* files are all intended to be strictly for debugging, making "real" differences stand out and unimportant stuff go away. 19:53
Or, if you like, note in a comment that it works either way.
As long as it is clear to our future selves that just because there is HLL scaffolding, it doesn't suddenly break the namespace handling. 19:54
NotFound Well, the main problem is that I don't know what interp['namespace'; 1] is supposed to be 19:55
japhb caller's namespace
(first caller)'s namespace, to be specific.
NotFound Urgh. 19:56
japhb Yeah, the API is ugly. I just use it, I didn't design it, I promise. ;-)
jhorwitz cotto: nm, i saw your e-mail
NotFound Yes, and the doc of the function is clear. Then yes, I think it does the same with or without parameter in the example. 19:58
19:58 integral joined 19:59 tomyan joined
NotFound japhb: fixed 20:04
japhb NotFound: thanks
NotFound: OK, so what's the current status? You believe that if lua if broken for fperrad, it must be fixed so that the current code works, and #853 should be rejected? 20:06
s/if broken/is broken/
dalek rrot: r40172 | NotFound++ | trunk/examples/opengl/static-triangle-hll.pir:
[examples] avoid unneeded namespace argument in import gl functions
 20:07
japhb s/current code/current OpenGL_funcs loading code/
sheesh
NotFound I think the fix is wrong, it uses hardcoded paths that doesn't woork with an installed parrot. 20:09
It can work if you use the compiled .pbc, but looking at the lua code it seems to be using the pir 20:10
So I'm confused.
20:12 mokurai joined
treed Coke: I commented on your bug with a PIR snippet that works around the bug, but which shows that your bug isn't the same as my problem. 20:12
NotFound Anyway, lua has no fixed release AFAIK, so if it is broken with 1.4 release, it can be fixed later. 20:13
Is there any other HLL that uses the opengl library? 20:14
treed Cardinal doesn't.
NotFound That's the bad part, I don't like very much the idea of breaking the only usage. 20:15
japhb NotFound: Rakudo, of course. 20:16
Well, Rakudo doesn't rewrap it -- you can just use it directly. Hence the .p6 examples. 20:17
NotFound That can be enough to do some tests... I'll take a look. 20:18
japhb lua rewraps apparently in order to exactly match the design of the OpenGL binding in "original" lua.
NotFound The problem seems to be that OSCON is interfering with people availability 20:19
japhb OSCON is a problem whenever I'm not attending. :-)
NotFound But if we all attend, who does the release? ;) 20:20
20:20 dukeleto joined
japhb Lightning Talk Live Release! 20:21
Tene japhb: you have access to cygwin, iirc? 20:22
japhb Tene: Not recently, no. 20:23
Tene Ah, nm.
Coke Tene: I do, on occasion.
Tene I added panels support to Curses.pir, and just kind of guessed the cygwin version of the name, based on google hints.
r40153
japhb Tene: any progress on the OpenGL::Math::Vec4 problem? Or was that put off for after the release? 20:24
Tene japhb: Um... no idea. I do have a counterexample that works fine for me, though.
japhb: Elementary.pir in github.com/tene/parrot-elementary/tree 20:25
treed Okay. 20:27
NotFound japhb: triangle.p6 works fine with parrot trunk installed.
treed So, I've done some further testing with implementing a cmp vtable sub.
Array doesn't like it, subclasses of array don't like it, but a new class will readily use it.
NotFound I'll try now with the revert to .include
treed (by don't like it, I mean that I get the message about no cmp with PP->I 20:28
dalek rrot: r40173 | dukeleto++ | trunk/CREDITS:
Adding myself to CREDITS
20:28 mokurai left
treed Bug with array, or am I doing something stupid? 20:28
dalek kudo: 3a2c5e2 | moritz++ | t/spectest.data:
add S02-names/identifier.t to spectest.data
 20:31
kudo: 0cdcc88 | (Bob Kuo)++ | README:
fixes typo in README file, 'git-format-patch' is actually 'git format-patch'
20:31 tomyan joined
NotFound What does 'use xxx:from<parrot>' in rakudo? load_bytecode 'xxx.pbc' ? 20:31
Tene NotFound: no 20:32
NotFound: :from<foo> does:
load_library foo\\n$P0 = compreg foo\\n $P0.'load_library'(xxx) 20:33
and then pulls symbols out of the return value of load_library and stuffs them into the namespace that called 'use'
japhb Tene: (re: Elementary.pir): Nice to see that NCI::Utils is getting some use by other libraries. Sucks that someone else is having to do the callback helper cha-cha. Although not exactly unexpected. Did you manage to make any design improvements for your callback helper library over the one I did for GLUT? 20:34
Tene japhb: I didn't know that you did one for GLUT
NotFound Tene: It uses xxx without adding any suffix? 20:35
Tene NotFound: it passes xx to the other language to interpret.
NotFound: as far as rakudo is concerned, it's magic. 20:36
are you askign what the 'parrot' language does with it?
NotFound I was not, but yes, I don't remenber what does trunk when there isno suufix specified. 20:37
Tene NotFound: if you want all the details, look at load_library in runtime/parrot/languages/parrot/parrot.pir 20:38
NotFound I think that deleting all pbc in all directories will be faster X-) 20:39
20:39 mokurai joined
Coke tene: can't build parrot on cygwin: no -lcrypt 20:39
NotFound Unfortunately rakudo is not helpful to decide about the question because it already depends on the parrot build tree. 20:41
japhb Tene: yes, config/gen/opengl.pm : gen_glut_callbacks() generates src/glut_callbacks.c , which compiles into libglutcb.so
pmichaud hello all
purl It's a crazy world, but hello to you too!
japhb o/ 20:42
20:46 dukeleto_ joined
NotFound Well, in doubt I'll keep the current status unless someone says a sound reason in time for the release. 20:46
20:51 payload joined 21:06 allison_perl joined 21:18 jrtayloriv joined 21:23 bacek joined 21:25 Whiteknight joined
dalek rdinal: fbdf8d1 | (Ted Reed)++ | (3 files):
Implement cmp vtable op for Array, and make the <=> operator use cmp; test included.
 21:54
bacek good morning #parrot 22:02
chromatic good moroning to you
allison good morning bacek
bacek I read "chromatic" instead of "purl"... 22:03
bacek need more coffee.
good morning allison and chromatic
dalek TT #866 created by bacek++: DEPRECATE (after review) some VTABLE functions 22:07
bacek allison: I have question about TT#784 (subclassing breaks multi dispatch) 22:10
What is "correct way to use MMD"?
(And my patch actually add MULTIs only for "math" VTABLEs. Not all of them) 22:11
dalek rrot: r40174 | bacek++ | trunk/DEPRECATED.pod:
Add more deprecations for Iterator and VTABLE
 22:12
allison bacek: aye, but not all PMCs should have multis even for the math vtables
bacek allison: Can you give example of such PMC? I don't quite understand it... 22:13
allison putting in a VTABLE function instead of multi dispatch is a valid optimization
so dispatch should go through the vtable functions primarily, allowing the PMC to decide if it should multiple dispatch
bacek I've put MULTI only for implemented VTABLE.
So, technically, it still use VTABLE for core PMCs. 22:14
And falling to MMD only if no candidates found.
Or I missed something obvious? 22:15
allison rereading the ticket 22:16
bacek Yay. Comment for "generate_multi_for_vtable" is wrong. 22:18
(copy-paste)--
allison bacek: it's mainly that generating multis for every math multi is a pretty broad and blunt way of working around a bug 22:20
bacek: I'd like to find and fix the bug, if possible
bacek allison: afaiu, registering VTABLE in MMD will work as expected. Not sure that is "fix for the bug". 22:22
allison bacek: I'm okay with applying it as a temporary fix, as long as we leave a ticketed todo attached to the code
the bug is that PIR inherited PMCs aren't appropriately traversing the inheritance hierarchy with MMD 22:23
bacek allison: ok. I'll do it tomorrow after release.
Whiteknight where is the dispatch code for VTABLE multis? I don't think I've ever seen it
allison Whiteknight: its the same code as all multis 22:24
Whiteknight oh, okay.
allison what a MULTI declaration in a C PMC does is insert an NCI in the global MMD namespace
Whiteknight ah, okay. that's the piece I was missing before 22:25
Whiteknight needs to spend more time looking at the generated code
allison I'm thinking I'd like to spend some intense time training people in parrot internals 22:26
bacek Whiteknight: nooooo. Don't do it! We need sane developers!
allison I was inspired by a meeting I'm in today
22:27 tomyan joined
allison talking about splitting out tasks from a role 22:27
Whiteknight what kind of training and what kinds of people? like existing developers or newbies?
allison existing developers
treed allison: I'd be interest in that.
bacek allison: Is "register appropriate VTABLEs in MMD directly without creating proxy " correct approach?
Whiteknight ditto, I'd be interested
treed Except I'm a newbie. :-)
And I'm not particularly a parrot coder. 22:28
Whiteknight There are plenty of parts of Parrot that I still only have a passing familiarity with
allison we've grown a lot in the past couple of years, and have a lot of people who have skills and interest in the internals
treed But internals are useful to know.
moritz I think the point of training is to make newbies into non-newbies
allison moritz: exactly
treed Gotta go to the market right now, though.
Anywhere I should pay attention for an announcement about this?
moritz I'm sure there'll be a mail on the list if there's a training session
allison and non-newbies into experts
bacek O! Can we deprecate "constant" PMCs? 22:29
and experts to gods!
allison bacek: we can refactor them, but we'll always need to be able to store constant PMCs
bacek allison: they are not "constant" in current parrot... 22:30
allison bacek: yes, it's a bit of a fiction
chromatic "constant" means different things to different people.
Whiteknight our handling of "constants" is pretty lousy throughout parrot, lots of premature optimizing
chromatic Here's a nickel; ask Bjarne if he's happy with the seven definitions of const you can't say on television.
allison bacek: because you have to be able to set a value on the constant before it's set in stone
bacek In parrot "zombie" is best description for "constant"...
allison bacek: zombies have benefits for optimizations 22:31
bacek And "Generational GC" is better :)
darbelo Probably related: some decidedly non-constant PMCs are allocated from the constant pool. 22:32
moritz wants generation zombies
chromatic Whiteknight's linked list GC scheme would work wonderfully with a constant PMC pool.
You almost get generational GC for free there.
allison bacek: generational gc is solving a similar problem
Whiteknight assuming my linked list GC scheme had worked at all 22:33
bacek and implemented :)
chromatic Your implementation was wrong, but the scheme is solid.
jhorwitz allison: [offtopic] do you see any reason to include the Parrot_cx_* functions in the embedding API? a few they look like they *might* be useful, but i'm not too familiar. 22:34
pmichaud I have a list of Parrot functions that Rakudo currently uses/relies on
one moment
Whiteknight chromatic: I'll do better next time :) 22:35
jhorwitz pmichaud: that will help me validate what i have, thanks. :)
allison jhorwitz: the ones related to scheduling a task or setting a handler are useful, but a good bit of it is internal to the virtual machine
jhorwitz allison: ok, for now i'm going to list all of them just so we don't catch anyone by surprise -- all 3 of us who are embedding. :) 22:36
nopaste "pmichaud" at 209.133.114.31 pasted "Parrot functions that Rakudo uses" (31 lines) at nopaste.snit.ch/17338
jhorwitz i've already eliminated everything obvious
pmichaud I included MACROS in that list 22:37
22:37 rg joined
jhorwitz thanks! 22:37
allison jhorwitz: sounds good 22:39
jhorwitz i'm really not sure how useful Parrot_mmd_* would be in an embedded environment either.... we should be trying to hide those details. 22:40
pmichaud Rakudo uses it because it implements its own multidispatch on top of Parrots
Is PARROT_API really only intended for those things in embedded contexts? Or is it intended to decorate any function that might be relied upon by things on top of Parrot? 22:42
bacek $dayjob time
see you
chromatic The former, I think. 22:43
pmichaud Is there a marker for things that HLL's might rely upon? 22:44
jhorwitz i think it warrants more discussion, b/c it's unclear from everything i've read. we have PARROT_EXPORT, which explicitly exports functions. PARROT_API should also implement that behavior, but be reserved for the API exported by libparrot.
allison jhorwitz: yes, the code should just be using the regular dispatch, which does multiple dispatch behind the scenes when needed 22:45
pmichaud I don't understand. 22:46
jhorwitz me or allison?
pmichaud allison, mostly, but a little of what you wrote as well.
allison in the current plan, PARROT_API is not functionality, it's documentation 22:47
pmichaud right.
I understand that part.
My question is whether HLL's should limit themselves to using functions marked PARROT_API
allison (commenting on jhorwitz's discussion point) PARROT_EXPORT is essentially "what various platforms require to be exported in order for parrot core to work" 22:49
chromatic Ideally.
pmichaud That understanding matches mine.
allison pmichaud: If an HLL needs a function that isn't marked as PARROT_API, that'd make it a serious candidate for being promoted to API status 22:50
pmichaud okay.
jhorwitz exactly
pmichaud My list then is the list of candidates :-)
jhorwitz which is why i'm being very careful about some of these. :)
22:50 tetragon joined
pmichaud (my list is the list of candidates that Rakudo would currently be wanting/needing) 22:50
jhorwitz pmichaud: the MMD ones are the only ones i don't have already, so i'll push those back on the list for now. 22:51
pmichaud Okay, great.
jhorwitz we're going to have a lot of functions at first. a good chunk of them are the Parrot_PMC_* functions though.... 22:53
pmichaud I'm fine if they _don't_ make it onto the list, as long as Rakudo has a way of handling its MMD needs :-)
(in the long run) 22:54
allison aye, we'll need to do some cleanup/refactoring. Establishing the current practices is a hugely valuable first step. 22:55
jhorwitz right, it would be ideal if we didn't have to expose these. but in this case, i defer to people who actually use those functions. :)
ok, dinner awaits. i'll fix up the docs later tonight in time for the release. 22:56
pmichaud jhorwitz++ # excellent work
moritz will try to allocate some time tomorrow for a fake Rakudo release 22:57
pmichaud moritz++ # that would be excellent
22:58 jrtayloriv joined
Whiteknight urg, there isn't an strtoll function on Windows 23:03
well, not in MSVC anyway
23:10 theory_ joined
cotto allison, ping 23:11
dalek a: 517eef4 | fperrad++ | src/lib/ (5 files):
load_bytecode PBC libraries instead of PIR

close TT #853		 23:16
a: 705f084 | fperrad++ | src/lib/ (2 files):
use the opcode 'iter' instead of a instantiation of 'Iterator'
23:21 kid51 joined
allison cotto: pong 23:31
cotto allison, is TGE at the point where it won't get any changes other than bug fixes? i.e. should rt.perl.org/rt3/Ticket/Display.html?id=40002 and all tickets it refers to be rejected?
allison aye, thanks for the message (got it earlier), was working on a reply 23:32
the answer is yes, PCT is the succesor to TGE
so most issues were resolved with PCT
(that is, PCT was specifically developed to address many of those) 23:33
cotto Great.
allison but, we do need to check the issues and make sure PCT addressed them
cotto ok 23:34
cotto just got a copy of Real World Haskell 23:38
mikehh cotto: excellent 23:39
23:39 patspam joined
dalek TT #860 closed by jkeenan++: Failed Pod extraction halts the build process 23:40
pmichaud if someone can summarize the issues from the tge tickets, I'll indicate whether they've been addressed or are likely to be addressed soon in pct
cotto Zak++ for recommending it and Learn You A Haskell 23:42
darbelo wonders if implementing haskell in parrot and using it to run pugs counts as 'having two perl 6 implementations'
cotto pmichaud, they're summarized in the first post of that rt I mentioned
cotto wonders about darbelo
pmichaud cotto: excellent. RT #40002?
cotto yup 23:43
pmichaud it's on my todo list. If you want to assigne the ticket to me, that might help also :-)
kid51 TGE: rt.perl.org/rt3/Ticket/Display.html?id=39913 23:44
cotto pmichaud, done
kid51 TGE: rt.perl.org/rt3/Ticket/Display.html?id=40002 (which u already have)
TGE: rt.perl.org/rt3/Ticket/Display.html?id=43757 23:45
pmichaud can assign those to me also
kid51 TGE: rt.perl.org/rt3/Ticket/Display.html?id=44979
pmichaud I'll even do a generic tge search a bit later
dalek rrot: r40175 | pmichaud++ | trunk/compilers/pge/PGE/Exp.pir:
[pge]: Failing to find a named regex should die directly instead of Null PMC
 23:48
lscript: 5eb1e9e | fperrad++ | src/runtime/wmlslang.pir:
load_bytecode PBC libraries instead of PIR
cotto Does anyone else pronounce PGE as "piggy"?
kid51 pmichaud: You should now have 3 additional RTs. 23:52
Whiteknight is moving over to win64 to do some final pre-release tests and update PLATFORMS 23:55
pmichaud kid51++ # thanks 23:56
(I think :-)

Powered by Cro and the Raku® Programming Language.

Please report any issues / comments / feature requests as an issue on App::Raku::Log.

Thank you!