Parrot 1.0 Released | parrot.org | 380 RTs left!
Set by moderator on 28 March 2009.
pmichaud attempting to do a svn checkout or update on the parrot repo gives me svn: Can't find a temporary directory: Internal error 00:17
now it's just hanging 00:24
pmichaud wanders off for a while. 00:39
dalek rrot: r37912 | jkeenan++ | branches/install_tools (4 files):
Add tests to mock the Cygwin-specific cases in Parrot::Install::install_files().
00:45
kid51 pmichaud: I was able to do 'svn ci'. Are you still experiencing problems? 00:46
01:17 kid51 joined
pmichaud seems to be working now, yes. THanks. 01:29
01:39 xinming joined
dalek rrot: r37913 | jkeenan++ | branches/install_tools/lib/Parrot/Install.pm:
Rename a variable to distinguish among various 'meta's.
01:54
rrot: r37914 | jkeenan++ | branches/install_tools/t/tools/install (2 files):
Add a few tests and specify exact number of tests in the plan.
02:23 amoc joined 02:39 davidfetter joined 02:41 janus joined 03:35 akl left
dalek kudo: 7dc65fd | pmichaud++ | build/Makefile.in:
Re-order Makefile targets into slightly more logical sense.
04:20
shorten dalek's url is at xrl.us/benwk5
eternaleye If I embed Parrot in another program, is there a way for that program to have Parrot parse input in some language and then access the AST produced? 04:21
04:38 iblechbot joined
dalek rrot: r37915 | allison++ | trunk/src/pmc/hash.pmc:
[cage] Fix comment, as Hash PMC does use the 'data' pointer now.
04:50
04:53 dduncan left 05:03 particle1 joined 05:33 Patterner joined 06:11 tuxdna joined
szabgab @tell chromatic trac.parrot.org/parrot/ticket/537 06:49
06:55 barney joined 07:49 masak joined 07:58 iblechbot joined 08:48 particle1 joined 09:29 TiMBuS joined 11:07 tuxdna left 11:25 barney joined 11:28 particle2 joined 12:06 iblechbot joined
dalek rrot: r37916 | fperrad++ | trunk/tools/dev/fetch_languages.pl:
[languages] add Cardinal
13:35
tracwiki: v88 | fperrad++ | Languages 13:39
tracwiki: update location of Cardinal
tracwiki: trac.parrot.org/parrot/wiki/Langua...ction=diff
shorten dalek's url is at xrl.us/benxab
13:41 Khisanth joined 14:04 Santtu joined 14:15 kid51 joined
dalek rkdown: c0a93c7 | (Francois Perrad)++ | t/MarkdownTest.t:
[spectest] list todo & skip tests
14:30
shorten dalek's url is at xrl.us/benxdk
14:48 bsdz joined
cotto Who's James Michael DuPont? 14:49
14:57 jan joined 15:01 Psyche^ joined 15:34 amoc joined
nopaste "bsdz" at 78.105.78.157 pasted "NQP list methods on my/our scoped vars" (9 lines) at nopaste.snit.ch/16130 15:42
bsdz hi all, just wondering if anyone can look at my nopaste. i'm wondering if it's a bug in nqp?
using "our" instead of "my" stops push from putting anything on a list. 15:44
pmichaud @mylist doesn't actually create a list -- it just declares a variable.
You really need our @mylist := (something) to create the list 15:45
i.e., NQP doesn't auto-vivify anything.
bsdz right i thought that might be the case. however, should i expect my_list and our_list to be different (in terms of contents)?
pmichaud yes, they're different variables. 15:46
bsdz yes i got that. but if i declare both and push 'some_data' on to each i would expect to see that in string in both lists. treating each individually 15:47
pmichaud sure, it should do that.
bsdz really?
pmichaud I guess I don't understand what you think it's not doing. 15:48
bsdz so if i push something onto @our_list, i won't see it but if i do it to @my_list i will?
pmichaud I don't understand what you mean by "i won't see it"
bsdz can you run the nopaste? 15:49
pmichaud no, because @our_list and @my_list aren't initialized.
bsdz yes but my_list gets populated and our_list doesn't
runs on my nqp btw
pmichaud oh.... perhaps my variables are being autovivified. I haven't looked. 15:50
anyway, the problem with the current script is that @our_list isn't being initialized.
bsdz how do i initialize it? if i add ":= ()" i get a "no result object" 15:51
pmichaud looking
bsdz thanks
pmichaud I'm not sure there's a way to do it strictly within nqp at the moment.
but I can fix that today.
bsdz that would be great. i guess i could vivify it manually for now 15:52
pmichaud that's what most scripts do.
bsdz it's just strange because "my @" looks like it autovivifies whereas "our @" doesn't 15:53
pmichaud yes, I think that might be the case right now.
bsdz thanks pmichaud - gotta reboot - c ya 15:56
15:59 Psyche^ joined
dalek kudo: 078012a | pmichaud++ | docs/spectest-progress.csv:
spectest-progress.csv update: 349 files, 8436 passing, 0 failing
16:05
shorten dalek's url is at xrl.us/benxj5
16:09 Andy joined 16:11 bsdz joined 16:15 frzntoz joined 16:18 frzntoz left 16:22 Theory joined
dalek rrot: r37917 | rurban++ | trunk/ports/cygwin (7 files):
[ports] cygwin 1.0.0-1 test candidate
16:26
16:48 Theory joined 17:01 Theory_ joined 17:12 Theory joined, mberends joined 17:25 AndyA joined 18:02 rdice joined 18:23 tetragon joined 18:37 eternaleye joined 19:05 szabgab joined 19:15 gryphon joined 19:23 eternaleye joined
Infinoid cotto: ping 19:24
msg cotto I noticed your TT #519, which I will have to rework my TT #18 patches for. I know x86 assembly rather well, but I don't know JIT at all. Is there anything I can do to help get your #519 patch working? 19:30
purl Message for cotto stored.
20:23 rdice joined 20:29 bsdz joined 20:33 ^conner joined
^conner morning all 20:34
purl afternoon, ^conner
20:38 Theory joined
^conner just out of curiousity, is parrot going to stay with svn for the long haul or is the a git conversion on the horizon? 20:42
moritz I don't there's anything planned for the near future 20:44
parrot switch svn hosts a few months ago; I don't think another switch will come along soon
Infinoid I suggested git once, they shot me down. Doesn't sound like it will change soon :) 20:45
That said, there are some of us using git on the client side anyway. 20:50
moritz is there something like a semi-offical git mirror somewhere? 20:52
Infinoid My git-svn bounce tree is on a decent server and has public read-only access; it's the best I know of
moritz URL? 20:53
Infinoid barney mentions it in his "leaving the nest" transitioning-history-to-github guide
git://squawk.glines.org/parrot-trunk/
It's updated every 10 minutes. I think the read-only bit limits its usefulness to anyone except me, tho 20:54
bsdz hi, any one might be able to help. i need to loop through a ResizablePMCArray in a Capture using NQP?
moritz what do you mean by "in a Capture"? 20:56
bsdz it's what gets created by a method in my actions.pm
moritz Infinoid: if there's an easy way to conince 'git svn dcommit' to push to parrot.org then it's probably rather useful 20:57
Infinoid There isn't. And all commits would be under my username
moritz meh
Infinoid Oh, you mean for local clones
moritz yes
^conner hmm, i havent' tried git as a svn client yet 20:58
Infinoid There's a clause in .git/config that needs adding, and some other metadata I haven't figured out yet
^conner I worked with svk for a while but I sort of drifted away from it as parrot was the only svn project I was using
at $dayjob, I just finally got the ok to convert from CVS -> SVN
Infinoid git svn is better than svn, but still pretty nasty compared to native git
^conner I've discovered that I rather dislike merging with SVN
Infinoid Yes, merging sucks. That's why I was pushing for git 20:59
^conner I don't understand why the svn folks can't just rip off the git merge tool
moritz bsdz: somehow I still don't understand what exactly you're doing... care to paste a short example? (I can't promise to help, I'm not all too deep into nqp) 21:00
^conner git would have a mindset too far for our internal users, it took years just to get them up to speed on CVS
Infinoid yeah, we have the same problem here.
^conner you start talking about distributed version control and heads explode
Infinoid ^conner: its not a matter of the svn merge tool sucking, it's a matter of it not having enough information to do a really good job
bsdz moritz: yes if you don't mind. i'm trying to figure out to boil it down though. i can nopaste a _dumper of my object? 21:01
^conner Infinoid, I think it's rather more retarded then that, even with a specified branch point, I got merges yesterday that put the ENTIRE branch version of the file as the left half, and the ENTIRE head version as the left half of the merge 21:02
moritz bsdz: sure
Infinoid: it seems that --commit-url might help, I'll try that now...
^conner Infinoid, I got better results with diff and patch
Infinoid ^conner: well, it was cautious and backed out the partial merge. having some kdiff3 integration would help with that 21:04
moritz: there's a database it maintains in .git/svn/git-svn/ (currently 7M in my tree) which it seems to need to enumerate the entire svn history to reproduce 21:05
^conner Infinoid, well you can at least tell it to use meld/kdiff3
Infinoid The simplest solution might be to just autocreate tarballs of the checkout.
Yeah. But having more information would solve some problems in my experience 21:06
^conner Infinoid, I ended the day yesterday with `emerge kdiff3` and went home ;)
Infinoid heh
moritz I think I should try that with a different repo than the parrot one first ;-)
Infinoid ^conner: What seems to happen in parrot land is, someone works on a branch for a while, merging trunk changes into the branch N times, and then does one final merge of trunk changes into the branch, before trying to merge the branch into trunk 21:07
nopaste "bsdz" at 78.105.78.157 pasted "loop through array in capture" (13 lines) at nopaste.snit.ch/16134
^conner Infinoid, that's exactly model we use at $dayjob, it just worked better under CVS
Infinoid Everything works fine until that last merge. Which is infuriating to me, because in git land that isn't even a commit... you've already *done* the merge, so it's just a bump of the master ref
Anyway, that has failed catastrophically for us at least twice so far. 21:08
^conner I believe it
Infinoid it seems to happen more often when files are renamed or removed, so there might yet still be a quick fix for it 21:09
^conner like I said early, you get better results getting a diff from head against the branch point
moritz bsdz: usually you don't try to iterate over match trees, but generate ASTs from them on-the-fly via action methods...
Infinoid Yeah, merging trunk changes into the branch makes that more difficult, but that's what I end up doing
moritz bsdz: that said if you want to access what's stored under the <ident>, you can use $/<ident>
Infinoid or just taking the merged tree and a fresh trunk checkout, and just rsyncing everything in :)
^conner Infinoid, agreed, it seems retarded but the model seems to need to be head->branch; resolve issues; branch->head 21:10
bsdz hmm , i admit haven't tried that. keep getting problems about "get_iter() not implemented in class 'Capture'" .. i'll give it a go
moritz if you want to iterate over positional captures, you can use 'for @($/) { ... }' 21:11
bsdz look familiar. think i'll give it another blast. dealing with a hll that returns parameters by declaring them in its function signatures. 21:13
^conner building parrot for the first time in like 6mo here... wow... is imcc still in use?
moritz ^conner: yes
^conner moritz, I thought the plan was to dump it? 21:14
moritz ^conner: yes. somebody[tm] just has to get pirc up and going, and then make all tests pass, and replace imcc... sooo simple :/ 21:15
^conner moritz, oh ya, no problemo ;) 21:17
this compiler warning bothers me: compilers/imcc/pbc.c:182: warning: 'old_blocks' declared 'static' but never defined 21:18
ya, old_blocks has a prototype declared twice 21:20
bacek good morning 21:27
moritz: there is another solution for imcc. Implement generating PBC from POST, implement PIR in PCT and use imcc for bootstrapping only :) 21:31
moritz bacek: yes, but maintaining a bootstrapped compiler means keeping byte code in the repository and very little fun when hunting bugs 21:32
bacek moritz: keeping bytecode in repo isn't mandatory. 21:34
^conner bacek, ya, only nessicary for the distrib versions
bacek, but it does mean that to start dev for a clean system you first have to install a dist version
bacek we can keep old imcc. 21:35
And probably reduce it size and features to bare minimum 21:36
^conner bacek, i don't think anyone wants to maintain lex/yacc unless they have too
bacek ^conner: pirc is still yac/lex based.
^conner shudders 21:37
i thought it was supposed to get done in PGE?
bacek ^conner: no. 21:38
implementing PIR in PCT is "my" idea :) 21:40
Infinoid: around? 21:43
bsdz moritz: hi, you still around? i've a boiled down pct elaborating my orig question 22:23
moritz bsdz: yes (but not for too long, I fear)
bsdz i understand - getting late here too. i'll nopaste . if you';ve time please take a look :) 22:24
nopaste "bsdz" at 78.105.78.157 pasted "looping through captures eg" (275 lines) at nopaste.snit.ch/16135 22:25
bsdz it's a hack based on complete_workflow.t. you need to prove t\\compilers\\pct\\bstest.t to run it 22:26
moritz bsdz: have you tried 'for @( $<return_identifier_list><identifier> ) { ... }' 22:29
bsdz yes get a capture has no get_integer i think 22:30
err no actually no, i get an empty array 22:32
that was the otehr weird thing
moritz what I don't understand is that 'for $<func_sig> { ... }' seems to work without trouble 22:39
bsdz i think i need to get out more on sundays. seem to have spent half a day on this! ;-) 22:40
moritz ask when jonathan or pmichaud are around, I'm sure they can help you much easier 22:41
bsdz thanks moritz. i'll keep a ping out for them :)
moritz wait, why do you want to loop over the identifiers? you already diid that in 'method return_identifier_list', didn't you? 22:44
anyway, off to bed now 22:45
bsdz good night 22:46
22:50 kid51 joined
Infinoid bacek: I am now, hi! 22:55
bacek Infinoid: good morning :) I start working on PBC's PMCs. github.com/bacek/parrot/tree/packfile_revamp 23:08
First steps - filling all paddings with zeroes to simplify binary comparisons constructed pbc with preconstructed. 23:10
Infinoid bacek: cool! I've asked jonathan to update PDD13 with his (relatively) recent changes to annotations. Until he does that, that part of the PBC PMC spec might not be fully up to date, but the rest of it is fairly accurate
That makes sense
I'm not sure I've ever used 3 acronyms in the same line before. That hurt. 23:11
bacek And I want to replace Packfile.get_string/set_string with 'pack'/'unpack' to be closer to PackFile functions
Infinoid You can add aliases, I guess, but get_string and set_string are what you get when you do $S0 = $P0 and $P0 = $S0 in PIR 23:12
or maybe it's set_string_native, I forget.
bacek "Write access to PBC from PMC will help to implement PIR compiler in PCT" :)
Infinoid That's a great goal, bacek++
bacek Infinoid: (strings) yes, they are. But for Packfile's strings has very different semantics. 23:13
Infinoid How so? It's still parsing and stringification, I think.
bacek For example PGE uses explicit functions instead of "set_string". 23:14
And it still "parsing" :)
Infinoid Well, it's a grammar engine, you can expect everything to be upside down and bizarre :)
bacek (And there is 4 acronyms in single sentence :)
Infinoid claps 23:15
bacek leaves in Australia. I can deal with "upside down" :)
Infinoid Better you than me, then
bacek :)
My english especially bad at the mornings... Need another coffee and commute to the work. 23:16
Infinoid No worries
Anyway, if alias methods make things more readable for your PCT work, they might be helpful elsewhere as well 23:17