Parrot 2.6.0 | parrot.org Log: irclog.perlgeek.de/parrot/today | Nopaste: nopaste.snit.ch:8001 | fix 'make html' (talk to Coke), merge gc_* branches, fix/replace/optimize hashing
Set by moderator on 3 August 2010.
00:08 lucian joined 00:25 LoganLK joined
cotto_work dukeleto: ping 00:27
particle osuosl will install smolder on the parrot.org vm 00:29
Paul_the_Greek If I make corevm, is it then more-or-less safe to make test_core instead of make test? 00:30
cotto_work msg dukeleto What tool do you use to migrate a svn repo to git and how could you get a map between svn revision numbers and git hashes? 00:37
purl Message for dukeleto stored.
mikehh Paul_the_Greek: the sequence is make corevm, make coretest, you can then do make or make world, make test 00:40
Paul_the_Greek Got it, thanks. 00:42
dafrito cotto_work: If you're looking to make a working copy, git svn clone -s svn.parrot.org/parrot (this has taken a -long- time for me, though)
cotto_work I'm not concerned with getting a working copy (though using the tarball on the wiki would be faster) as much as figuring out how to write a trac plugin that'll see r12345 and know how to map that to a link to (e.g.) a specific commit on a github repo 00:45
mikehh Paul_the_Greek: most do not bother with make corevm/make coretest, however I test that to determine coretest works with corevm as sometimes this breaks 00:46
Paul_the_Greek I'm working on memory management, so I can test it with only the core. After I think it's all set, then I can test everything. 00:47
00:47 plobsing joined
mikehh i.e. a test is written that requires some later parts of the build but runs in coretest 00:47
Paul_the_Greek Is there a preferred #if paradigm for commenting out code? 00:49
mikehh you need to do a make clean/realclean before running make corevm/make coretest
cotto_work Paul_the_Greek: are you planning on committing something that comments out a chunk of code?
Paul_the_Greek Oh, I didn't know that. What is the difference between them?
cotto_work realclean? 00:50
purl realclean is, like, sometimes needed because of parrot's bad dependency checking
mikehh Paul_the_Greek: yes #if 0, but it is not generally recommewnded
Paul_the_Greek No, I really will delete the code before proving the patch. But in the meantime, just in case ...
#if 0 is what I used.
clean? 00:51
purl clean is fine, just don't fucking hide every goddamn thing in sight, especially the goddamn soap.
plobsing msg purl msgs
purl Message for purl stored.
cotto_work messages?
purl To access purl's messages, msg me with the word "messages".
Paul_the_Greek msg purl messages
purl Message for purl stored.
Paul_the_Greek Okay, I'm just confusing myself.
plobsing ETOOMANYMSGS 00:52
seen chromatic?
purl chromatic was last seen on #parrot 18 hours, 24 minutes and 18 seconds ago, saying: msg plobsing Can those turn into lazy thunks?
Paul_the_Greek seen paul_the_greek 00:53
purl paul_the_greek was last seen on #parrot 0 seconds ago, saying: seen paul_the_greek
Paul_the_Greek Every group needs a secretary.
mikehh Paul_the_Greek: I always do a make realclean, configure with --test, and then the other builds and tests 00:54
plobsing msg chromatic: To know to do that (and how), you'd need to have decoded them. The real problem is with serialization/deserialization having side-effects in the first place. Serialization should be dumb (and certainly shouldn't get "smarter").
purl Message for chromatic stored.
sorear (does Paul_the_Greek have a CLA yet?)
kid51_learn_rakudo Paul_the_Greek: 'make realclean' undoes the effect of both 'perl Configure.pl' and 'make'
Paul_the_Greek Why do I want to undo the configure?
kid51_learn_rakudo Paul_the_Greek: 'make clean' only undoes the effect of 'make'; configuration is left intact, hence the Makefiles remain
Paul_the_Greek: Some time you do, some times you don't. 00:55
Paul_the_Greek Without the configure, how does it even know how to perl or compile something?
kid51 If someone has changed something in the configuration system -- which, I'm glad to say, happens rarely these days -- ...
Paul_the_Greek sorear: What is a CLA? 00:56
purl a CLA is Contributor License Agreement or www.perlfoundation.org/contributor_..._agreement or www.perlfoundation.org/attachment/legal/cla.pdf or www.parrot.org/foundation/legal or www.parrot.org/files/parrot_cla.pdf
kid51 you will need to reconfigure.
Paul_the_Greek Aha.
kid51 More frequently, someone will change config/gen/makefiles/root.in, the template for Makefiles. Then it's also wise to reconfigure, though IIRC there is a 'make reconfigure' (YMMV)
Paul_the_Greek Ooh, a license agreement.
kid51 But let's say you're hacking away in a branch or in your own github ... 00:57
... and all your changes are under directories like src/ ...
... then you probably don't need to change the Makefile, but you *do* need to blow away all the files generated from the C source, including the parrot executable.
Paul_the_Greek Why wouldn't make corevm do that anyway? 00:58
00:58 dduncan joined
kid51 Examine config/gen/makefiles/root.in and the other .in templates -- or examine the Makefiles -- to learn more about different targets. 00:58
'make corevm' is a more specialized bird
Paul_the_Greek Examine a makefile? Okay, hang on, going to get a large quantity of rum ... 00:59
kid51 It refrains from linking many libraries
nopaste "plobsing" at 192.168.1.3 pasted "[PATCH] buggy lazy constant table" (4656 lines) at nopaste.snit.ch/22602
mikehh Paul_the_Greek: it doesn't - and if you don't run make clean there would be file left over from make/make world
files
Paul_the_Greek Okay, make clean first, always.
mikehh like .o etc
kid51 make corevm is something that of use to people hacking on the Parrot core ... 01:00
Paul_the_Greek Can I sign the CLA by just editing the pdf file and then email it?
kid51 ... and make coretest exists to test what they've been hacking on ...
Paul_the_Greek Yes, I'm working on memory management. 01:01
kid51 ... but it's not intended for use by, say, people developing HLLs on top of Parrot ...
01:01 dduncan left
kid51 ... for those people, ie. most of Parrot's *users*, 'make test' is more critical 01:01
Paul_the_Greek Yes, that makes sense. 01:02
kid51 So, unless mikehh corrects me, I would say that 'make corevm; make coretest; make clean' is a good cycle for you
plobsing msg chromatic: Lazy constant table would be an easy win if packfiles and certain freeze-thaw vtables were more well behaved in certain areas. I'm done with this direction for now (too big a project for the benefits). Here's what I have so far nopaste.snit.ch/22602 01:03
purl Message for chromatic stored.
mikehh I generally run make realclean, configure (with --test and some other options) then make manifesat_tests, make codetest, make corevm, make coretest 01:04
then make world, make test, (maybe make distro_tests) and the make fulltest 01:05
kid51 mikehh: But that's when you're doing your comprehensive smoke testing. I think that's overkill for Paul's needs
darbelo plobsing: How well-behaved and in what areas?
mikehh sure, but that's what I generally run, mind you I usually insert a make realclean, make world (logging) and make smoke, before fulltest. 01:06
kid51 Paul_the_Greek: IIRC, you have to postal mail or fax the CLA -- but that may have changed
Paul_the_Greek Yes, the CLA pdf is not editable. I'll print and mail. 01:07
mikehh I think the email option is acceptable if you get your signature on it (the PDF that is)
kid51 mikehh: But for his minute-to-minute development cycle, I think 'make corevm; make coretest; make clean' would be best. Once he's ready to post a patch, then 'make test' is needed
Paul_the_Greek By the way, I have an industrial-strength color laser printer here. If anyone needs a large document printed, I can do it and ship it to you. 01:08
mikehh don't forget make codetest
plobsing darbelo: well, we re-use packfile directories (decoded from PBC) in ways we shouldn't. We depend on deserialization of a number of different types to set up our runtime environment properly (ParrotInterpreter is the worst offender)
The inner workings of some things are simply too magical 01:11
Paul_the_Greek That's when you toss it and rewrite it with your own personally acceptable magic.
kid51 reads Parrot CLA 01:12
If you can make a PDF of the *signed* CLA, you can send PDF as email attachment. 01:13
Otherwise, fax or postal mail.
plobsing Paul_the_Greek: No! We're all out of the good magic. We've been using the bargain brand for years, and look where it got us.
Paul_the_Greek Kid51: I could do that, but I've already stamped, addressed, and filled the envelope. 01:15
I have a little fresh magic here, and then I can fall back on conjuring. 01:16
dalek rrot: r48340 | darbelo++ | branches/unshared_buffers (20 files):
Sync with trunk.
 01:17
Paul_the_Greek Is there a Parrot Conference?
kid51 Paul_the_Greek: Regrettably, no. We have Online Developer Summits 2 or 4 times a year. 01:18
Paul_the_Greek Where you all spend two days in this chatroom? 01:19
kid51 is a big proponent of F2F hacking
No, it's a 3-hour session
There is weekly conference in #parrotsketch at 1830 UTC
correction
purl engage brain. then fingers.
kid51 2030 UTC (4:30 pm US EDT, IIRC) 01:20
Paul_the_Greek Which day?
purl rumour has it Which day is it? I might be busy that day
kid51 All disputes are resolved by purl
Tuesday
And we release 3rd Tuesday each month
Paul_the_Greek Tuesday, 4:30 PM. Why in #parrotsketch instead of here?
kid51 Because have lots of pointless chatter in this room at *all* times ;-) 01:21
And we need a channel we can log as if we were taking minutes
In #parrotsketch, the meeting is called to order and is later adjourned 01:22
Paul_the_Greek Right, a clean channel. That makes sense.
Let me add a sticky to my desktop ...
kid51 In #parrot, either the meeting never stops ... or the concept of order has no meaning
See the irclog thingee in the topic?
Just substitute parrotsketch for parrot and you can read dialog at recent meetings
I'm at $job at that time, so I'm almost never in parrotsketch 01:23
... which those in the know abbreviate to 'ps'
01:23 rurban joined 01:24 Essobi joined
Paul_the_Greek I'm pleased to see all these wasted cycles being used by Parrot makes. 01:25
kthakore Paul_the_Greek: whut? 01:26
purl stuff.
plobsing If Parrot didn't tie up my CPU cycles, they might get wasted idling. 01:27
Paul_the_Greek A cycle unused is gone forever. Might as well use them to do Parrot makes.
The amazing thing is that make is using all 8 processors on my machine. I'm impressed.
sorear Paul_the_Greek: incorrect. a CPU has variable power consumption, and every cycle used directly translates into money you have to pay at the end of the month for electricity 01:28
unless electricity is on a fixed monthly rate in Greece?
Paul_the_Greek Nonetheless, the unused cycle is lost forever. 01:29
I don't mind a little higher electric bill, even here in Massachusetts.
That saying about the unused cycle had more punch in 1967. 01:30
So did the saying about the handy halfword of zeros.
And the saying about weight-lifting the ROM boards of the Meta4. 01:31
mikehh Paul_the_Greek: there is a log of the IRC at irclog.perlgeek.de/ you can check out some of the discussions there for #parrotsketch (and others)
Paul_the_Greek Oh good, I'll read the last discussion. 01:32
darbelo We call 'em the irclogs, most of the time.
Unless infrared clogs factor into the dicussion, since it'd be ambiguous. 01:33
Paul_the_Greek Well folks, I'm off to crash. Thanks for all your help. 01:35
01:38 rurban_ joined
dafrito cotto_work: I wrote a map-git-to-svn script, found at pastebin.com/g7AWvrSM Here's a snippet of the map it makes: pastebin.com/FWj2bTdZ 01:39
I matched 43000 of the 48000 revisions. I'm looking into where the rest went ;) 01:40
cotto shiny 01:50
dafrito++ 01:51
Once my trac test environment is alive, writing the plugin should be pretty simple. As far as I'm concerned, getting the map is the hard part. 01:53
dafrito definitely 01:56
cotto Hey Firefox, that 140M file ending in .tbz isn't HTML. Please don't try to render it. 02:00
02:00 dngor_ joined 02:11 workbench joined 02:12 slavorg joined
Austin I give up. What's the difference between +o and +oo and +ooo ? 02:21
Is there a Pink Floyd-bot or something?
cotto This sounds like a really nerdy riddle. 02:23
dafrito An o for each person, I think
cotto dafrito, does that script expect to be run from the root of a git-svn clone of parrot, and should I expect it to not do anything for a while? 02:24
dafrito cotto yes to both
cotto ok then
sorear opbots trust dafrito 02:25
slavorgn Ok
slavorg Ok
cotto dafrito, it's always nice to see a new nick here. What's your interest in Parrot? 02:26
dafrito \\o/
sorear opbots beleive p6eval
opbots believe p6eval
slavorgn Ok
slavorg Ok
dafrito I found parrot through the JVM language summit, and found it much more pleasant than the JVM 02:27
nopaste "kid51" at 192.168.1.3 pasted "nopaste is back" (1 line) at nopaste.snit.ch/22603 02:29
kid51 nopaste, status
purl seen nopaste?
purl nopaste was last seen on #parrot 32 seconds ago, saying: "kid51" at 192.168.1.3 pasted "nopaste is back" (1 line) at nopaste.snit.ch/22603
kid51 purl nopaste status? 02:30
purl kid51: bugger all, i dunno
kid51 purl status nopaste?
purl Since Thu Aug 5 20:55:17 2010, there have been 135 modifications and 136 questions. I have been awake for 1 day, 5 hours, 34 minutes, 55 seconds this session, and currently reference 853053 factoids. Addressing is in optional mode.
kid51 purl seen dalek?
purl dalek was last seen on #parrot 1 hours, 13 minutes and 25 seconds ago, saying: parrot: review: trac.parrot.org/parrot/changeset/48340/
02:39 khai joined 02:55 janus joined
dafrito cotto how many revisions did you match? 03:01
04:16 Andy joined 04:22 AndyA joined
plobsing seen NotFound 04:45
purl NotFound was last seen on #parrot 2 days, 18 hours, 20 minutes and 44 seconds ago, saying: Maybe we must make that parameter optional and provide a useful default. [Aug 4 10:24:19 2010]
05:17 desertm4x joined 05:28 bacek joined 06:34 bacek joined 06:40 robin-gvx joined 06:53 fperrad joined
cotto dafrito, none. The script did nothing for 12 minutes and then stopped without any output. 07:38
got it working 08:11
08:12 AzureStone joined
cotto Hmmm. Each svn revision seems to map to two git hashes. 08:27
sorear the tree and the commit?
cotto not sure 08:28
Austin cotto: looking at the script, it seems to be printing an attribute that is probably generated by the git-svn bridge (not a git builtin). Don't know if that helps. 08:43
08:49 AzureStone joined
cotto I'm using it on a git-svn clone 08:53
08:53 Essobi joined
Austin cotto: Sure. My point is that if you want to know what the two hashes are, you'll probably find them in the git-svn, as opposed to pure git, docs. 08:56
cotto gotcha 09:06
You're sure up late.
Austin Heh. Not any more. It's early now. 09:07
09:28 desertm4x joined
dalek kudo: 6ebba85 | pmichaud++ | src/Perl6/ (2 files):
Fix handling of $_ scoping in routines (broke due to $*IN_DECL change in
 09:36
09:36 ttbot joined 09:38 rurban_ joined 10:08 jjore_ joined
fperrad msg muixirt thanks for the report (see TT#1733) 10:15
purl Message for muixirt stored.
10:20 whiteknight joined 10:24 dmagnus_ joined
dalek rrot: r48341 | moritz++ | trunk/runtime/parrot/library/Digest/MD5.pir:
add parrot .HLL to Digest/MD5.pir
 10:58
11:03 dmagnus__ joined
dalek kudo: e569d66 | pmichaud++ | src/Perl6/Compiler/Signature.pm:
Remove "$_ is always first argument" from Perl6/Compiler/Signature.pm, fixing RT

I'm not entirely certain why it was being put as the first argument in the first place, but all of the spectests pass without it (and we now resolve the RT ticket).		 11:15
whiteknight in NQP, how do I get a reference to the protoobject of a class? 11:32
Austin Starting from where? 11:33
whiteknight I have a Class object
Austin Okay.
whiteknight in Kakapo::UnitTest::Loader.configure_suite
Austin And you know it's a P6object, not a PMC...
Have a look at Testcase.nqp // sub TEST_MAIN 11:35
The first couple lines go from a namespace to the name to the protoobject 11:36
You can get the namespace from the class.
(If it has one.) 11:37
$ckass,get_namespace() 11:40
yikes
$class.get_namespace()
Alternatively, call $class.WHAT() (or $object.WHAT() if you have an object). 11:44
(but pmc's won't support the p6object methods, I think) 11:49
whiteknight ok
I have t/Program.nqp running. It produces zero output, but doesn't have any errors 11:50
the rest of the tests are not so lucky
Austin whiteknight++
12:05 muixirt joined
whiteknight it's calling Suite.run(), but there is apparently nothing in @!members 12:13
12:36 ash__ joined 12:38 kid51 joined 12:39 ash_ joined
whiteknight pir::say("add_test Members: " ~ pir::elements__IP(@!members)); 12:50
pir::push__vPP(@!members, $test);
pir::say("add_test Members: " ~ pir::elements__IP(@!members));
the number of elements is zero before AND after the push
can anybody spell WTF? 12:51
Austin Try @!members.push: $test 12:52
I think you're maybe getting an rvalue op, because nqp doesn't know to store the result of ::push into the attribute. 12:53
whiteknight same result
Austin Yeah
whiteknight 0 elements before and after the push
Austin That was a stupid suggestion.
Compile it to pir, and nopaste it. 12:54
nopaste "Whiteknight" at 192.168.1.3 pasted "UnitTest::Suite.add_test in pir for Austin++" (39 lines) at nopaste.snit.ch/22611 12:56
Austin So if the attr is null, it autovivs and doesn't store. 12:57
whiteknight that sounds like a bug
Austin Sure.
But attr handling in nqp is a mess, 12:58
...
12:58 contingencyplan joined
Austin @!members := @!members 12:59
Then do the rest
whiteknight Austin++ that did it (and exposed new errors further down the rabbit hole) 13:02
Austin woo-hoo! new errors
whiteknight let me check in my progress so you can take a look at it 13:40
github.com/Whiteknight/kakapo/commi...9979b1b07b 13:42
13:44 mak joined
mak Hi 13:45
I asked this question on #perl6 13:46
After doing this gist.github.com/512792
rakudo refuses to build
whiteknight it built before?
mak yes, actually im trying to trace how rakudo executes things... by using print statements 13:47
gist.github.com/512785 this is the error
kthakore whiteknight: hi
purl hi, kthakore.
mak was unable to get an answer on #perl6 13:48
whiteknight hello kthakore
mak is say a valid pir statement?
whiteknight mak: when rakudo does't build, what is the error messag eyou get?
kthakore whiteknight: got some releases out. Have some time to make parrotSDL work ... again
mak gist.github.com/512785
whiteknight mak: yes, "say" is a valid PIR opcode and you're using it correctly it seems
mak: to satisfy curiosity, can you try changing "say" to "print"? 13:50
mak ok
nopaste "Austin" at 192.168.1.3 pasted "Using say opcode" (7 lines) at nopaste.snit.ch/22612 13:52
whiteknight bah, that's all I've got for today 14:25
or, at least for the morning 14:26
14:31 Andy joined 14:36 mak joined
mak sorry i got disconnected last time due to power cut here 14:37
but using print instead of say didnt help
is anybody there? 14:41
Austin mak: That code isn't broke, according to my parrot.
(the code you nopasted) 14:42
I suspect something is wrong in the surrounding text. Can you nopaste the whole file?
mak gist.github.com/512869 14:43
Removing the say or print builds rakudo perfectly 14:45
Austin mak: What happens if you load the hello world thing I nopasted above? 14:49
mak I cannot see the link
Austin nopaste.snit.ch/22612
Sorry, I forgot you moofed 14:50
This is weird, but what if you put the print/say statement above the commenb lines? 14:51
*comment
mak your hello world program worked, now checking by putting the say above the comment 14:53
Austin So we know you've got built-in say or print. 14:54
mak putting the say above the comment made no difference 14:58
Austin Well, I guess that's a relief
mak error:imcc:syntax error, unexpected IDENTIFIER, expecting $end ('from')
thats the error
pir files can have blank lines?? 15:00
arnsholt Sure they can
Austin What is the line number again? 15:01
mak I think Im missing something very fundamental here
Austin Yeah
Wait, is there a unix/dos issue?
Are you on windows?
mak like? I using ubuntu
Austin So there's not secretly a bunch of cr/lfs in the file? 15:02
mak Im using emacs
Austin Okay
If you delete the print statement inside main, the problem goes away, right?
mak yes 15:03
Austin What line# is the print statement in main? 15:04
228?
purl well, 228 is 12+1
mak precise error message gist.github.com/512882 15:06
Austin Sure, but that's the reported error location. Where's the print statement?
'cause to me it looks like 228, or 220-230 if you've been moving the comments around 15:07
mak 1 min
224
Austin What's your local language? 15:08
mak how does the line number matter
Austin Any chance of funny quotes?
mak kannada
Austin The line number matters because the error is reported 40+ lines away.
mak Im from India
Austin When you get a really far away error report, sometimes it's because you forgot to close quotes, or parens, or brackets. 15:09
Hmm. Try deleting (ctrl-k) the line with the print statement. Then copy/pase the line @266 or so with "say 'Copyright perl foundation'" 15:11
mak power goes again, on ups... can't hold on for long... will join again in a short while 15:12
kthakore svn head is broken 15:17
mak: hi from india!
mak: kaisa hey yaar! 15:18
paste.scsys.co.uk/47721
Austin: ^^ svn head
Austin What rev? 15:19
kthakore 48341
darbelo Have you tried doing a make realclean? 15:20
Austin Whew
I'm not at that rev...
kthakore darbelo: let me try again ...
Austin So I guess I'm safe..
kthakore Austin: hahaha ... I thought you were gonna help
:p
Austin doesn't do 'C' 15:21
kthakore Austin: oh well
kthakore is all about the C
:D
darbelo I have a succesfuly built r48341 here. That's why I suggested a realclean. 15:22
kid51 r48341 build okay for me 15:23
kthakore darbelo: doing it again ...
kid51 r48341 make test PASS on linux/i386 15:26
kthakore kid51: that is what I have 15:27
ok ... weighting
btw ... does anyone know if anyone has done a memory profile of parrot? 15:29
like ccmalloc ?
mak kthakore: hi, sorry there was no power
sorear We use a private allocator.
kthakore sorear: no I don't mean that 15:30
sorear: I was wondering if we have profiled parrot tests to see where most of the memmory is consumed?
sorear That's so easy we don'e even need to profile./ 15:31
init_pmc.
purl init_pmc is, like, called when a pmc is passed to init
sorear (or is it new_pmc? been a while)
kthakore sorear: I don't understand ...
oh you are saying init_pmc takes most memory?
but where ?
in init_pmc
sorear On the garbage collected heap.
Austin kthakore: You can't go more than a day or two on here without chromatic showing up burbling about shaving 0.011% off memory consumption. Check the irc logs for the last few days, there were some guys showing various profiling scripts. 15:32
mak Austin : putting the say on 269 worked, but putting the very same thing back again in the main didn't ... weird 15:33
gist.github.com/512902 15:34
Austin mak: what happens if you move the say to just before the .end in the main sub?
kthakore Austin: ok
15:39 Andy joined
mak This time a different error 15:42
Austin Well, that's progress. 15:43
mak gist.github.com/512910
Austin If we can change the error, we must be in the right place
mak :) what a bad start when learning something new
Austin Well, that's a run time, not compile time, error. 15:44
Try moving the 'say' to after the !fire-phasers(INIT) call. 15:45
mak ok
build speed is slow... a cycle of change-build-test takes a little time 15:46
Austin Yeah, I know. 15:47
mak same old error again gist.github.com/512914 15:50
Austin That's baffling 15:51
I could understand if it depended on some kind of init-foo.
Maybe it's the compreg? 15:52
Try moving down one more line (after the compreg)
My bad. I should have thought of that before.
compreg = register a compiler
It shouldn't have any effect - the compreg in question is a get, not a set. 15:53
But I guess it might...
Otherwise, I have no idea why it's failing.
mak #perl6 guys couldn't help either 15:54
Austin You probably want pmichaud or jnthn. I'm not 100% sure who else is elbow-deep in rakudo...
mak putting it after compreg didn't work 15:57
Austin Well, that's good. 15:58
At least the get/set thing is true.
mak There have been similar such commands in the log, gist.github.com/512925 16:00
there its -o src/gen/perl6.pbc 16:01
but in here its just -o per6.pbc
Austin Run this command, tail -n +222 xx.pir | head -10 | od -t x1z 16:02
except change xx.pir to the name of your pir file
mak gist.github.com/512926 16:03
Austin THat looks depressingly normal 16:06
Hmm 16:07
try head -10 src/gen/core.pir
mak gist.github.com/512932 16:08
Austin I'm thinking that first line looks wrong
TiMBuS Austin, i can confirm theres nothing odd about the file except the 'say' opcode being added
Austin timbus: Thanks 16:09
TiMBuS it's really strange
Austin Aha 16:10
!!
mak eureka???
purl eureka is not a good place to live, trust me or skud's web log analyser that uses MySQL to make it easy to drill down on information, at sourceforge.net/projects/eureka/ or as Skud says, a good idea with a piss-poor to non-existent implementation, please don't look.
Austin mak
mak yes
Austin The first run is outputting what you want, with -o to some other file.
mak didn't get you 16:11
Austin The second run is inputting that file, with the added line, and failing because the extra line is syntactically invalid as input
The first run of Cmpiler is emitting "Copyright perl foundation", but it is redirected by -o src/gen/core.pm 16:12
So the "Copyright ..." become line 1 of src/gen/core.pm
Then, the next time the makefile runs Compiler, the input file (src/gen/core.pm) contains a syntax error: "Copyright ..." at its first line.
TiMBuS i guessed that Compiler.pir was being preprocessed in some way, but thats a bit different to what i was expecting.. 16:13
Austin The .include directive @line 270 or so shows up as the error location.
EVen though it's really in a different file.
mak Then how come other say's in the Compiler are not throwing errors and only the ones in main?
Austin timbus: Compiler isn't being preprocessed. core.pm->core.pir is 16:14
They aren't being run.
mak ohhhhkk
Austin mak: They aren't being run, and they aren't errors.
Try this.
Replace the say in main with printerr
printerr 'Copyright ...' 16:15
MAybe add \\n
mak or something like ... say 'say "Copyright Perl Foundation"' 16:16
TiMBuS Austin, yeah that's what I meant. it was quite different to what I had expected, but now its totally obvious
Austin timbus: See here, third line uses >src/gen/core.pir gist.github.com/512925
mak: No, because first line is not inside correct structure. (sey has to be inside .sub/.end block). Maybe a comment instead : say '# Hello world' 16:17
16:18 tcurtis joined
Austin mak: Result will be first line of src/gen/core.pir if you do a print or say. Result should appear on console (stderr) if you do a printerr. 16:19
mak The whole point of putting those say's is to use them as kind of debug logs 16:20
Austin Heh
mak yes , printerr stuff works
just tested
:)
sorear clearly we need a 'note' opcode.
Austin That works, until you add >src/gen/core.pir on the outside of the script. Then your log message gets written to the wrong place.
:)
Austin ++
mak or maybe while using debug statements instead of printing it out on stdout .. print it out in a file 16:22
does parrot have IO stuff?
file IO i mean 16:23
Austin Yeah, there's dynamic ops (open) and FileHandle pmc type
mak Lots to learn... good I got this error... learn't a lot today 16:24
thanks Austin 16:27
Austin Happy to help, mak. I'm sorry it took so long.
mak Well I was just thinking ... In that case #perl6 folks must also use printerr's instead of say's 16:32
Compiler error messages must be use the printerr instead of say 16:34
16:35 robin-gvx joined 16:36 mak left
dalek rrot: r48342 | darbelo++ | branches/unshared_buffers/src/string/api.c:
Fix braino in chopn semantics.
 17:20
17:38 rurban_ joined 17:58 bacek joined 18:02 Andy joined
dalek kudo: fca2d32 | pmichaud++ | src/core/Any-list.pm:
Fix Whatever in .[*] and .{*} . Fixes RT #77072.
 18:40
18:52 plobsing joined 19:05 Casan joined 19:24 Andy joined 19:42 s1n joined 20:53 LoganLK joined 21:10 jsut joined 21:13 preflex joined 21:17 Paul_the_Greek joined
Paul_the_Greek Hey folks. 21:17
mikehh greetings, Paul_the_Greek 21:26
Paul_the_Greek A much more pleasant day here.
mikehh not too bad here, except night has falen 21:27
two l's
Paul_the_Greek Where are you?
mikehh Aberdeen, Scotland 21:28
Paul_the_Greek Ah, wonderful. My wife was just in England and France. We were in Scotland many years ago.
I'm in Massachusetts, USA.
21:30 bacek joined
Paul_the_Greek So when do you think the Lorito project is going to get really serious? 21:32
mikehh ah well. bacek is in Sydney where it is 7 32am tomorrow
Paul_the_Greek Is he the Lorito guy? 21:33
mikehh It's been under discussion for a while now, but we are going through the design process at the moment, with a couple of prototypes 21:34
cotto, chromatic
Paul_the_Greek What is being used to assemble it now? 21:35
mikehh I don't think it has got that far yet - there is a prototype at github.com/atrodo/lorito 21:39
just looking at some concepts
check out - wknight8111.blogspot.com/2010/07/lo...esign.html 21:42
cotto There are a couple prototypes but they're very experimental atm.
One of my weekend projects is to wrap my head around using CPS in an assembly language context. 21:43
mikehh cotto: what's that Continuation passing style? 21:46
Paul_the_Greek Is the ultimately a sophisticated macro assembler for Lorito, or is that the job of higher-level components? 21:48
mikehh my understanding at the moment is that lorito will be a sort of micro-architecture, with the current PIR riding on top of it 21:54
so that our current op codes will generate one or more lorito ops for each higher-level( current) ops 21:55
Paul_the_Greek So no one will write code in Lorito ops, but rather continue to write PIR code? 21:57
The PIR compiler will be changed to emit Lorito, then. 21:58
mikehh well probably even higher level NQP for example or ultimately Rakudo or Partcl or something like that 21:59
obviously we need to provide the infrastructure for that
cotto mikehh, that's correct. 22:00
I
mikehh I think that is where pir/PIRATE is supposed to come in 22:01
cotto 'd probably make both of us more confused if I tried to explain CPS. Wikipedia is a good starting place.
mikehh if you like Scheme or something
cotto needs to go
Paul_the_Greek I'm trying to get a handle on the Lorito assembly process because I think that might be something I could help with.
The current status of the assemblers is confusing. 22:02
mikehh I think initially each lorito op will effectively be a C function call or something like that 22:03
Paul_the_Greek Yes, that's the impression I get. 22:05
And if humans don't write Lorito, then no fancy macro assembler is needed. 22:06
IMCC appears to be a mudpile, while PIRC seems to be stagnant.
mikehh well we had IMCC, which has evolved to be almost incomprehensible, kjs started PIRC which has at this stage not evolved into our new assembler, pir/PIRATE seems to be a new option
Paul_the_Greek PIRATE? 22:07
purl PIRATE is a career that is washed up these days...and raping and pillaging just doesn't have a good reputation these days.... or www.eff.org/IP/DRM/piratead/CEA_ad.png or www.wired.com/politics/security/mag...li_pirates or PIR parser (and compiler) implemented in NQP github.com/bacek/pir or tweetphoto.com/28970505
mikehh one of our problems here is that we have only had one or two people working on IMCC or PIRC which is not a good bus number
Paul_the_Greek There is no code directory for PIRATE. It must be a separate project. 22:09
Ah, I found separate info on PIRATE. 22:10
mikehh there was some work on opsc by bacek & cotto, bacek then re-implemented pir (as PIRATRE) to take advantage of that, both these projects were a SORT OF LEAD-IN TO lORITO 22:12
pirate
dammit - hit the CAPS-LOCK by mistake 22:13
anyway compilers/opsc and pir/PIRATE are projects that should make implementation of Lorito easier 22:14
at the moment they use nqp-rx which could raise some bootstratpping problems but hey 22:16
Paul_the_Greek So maybe I could help with PIRATE. It looks like it's in need of help.
mikehh our main coding-robot on that - bacek - seems to have taken a work-related break recently, hopefully he's back 22:18
22:19 kid51 joined
Paul_the_Greek Who is Tyler Leslie Curtis? His name is all over the place. 22:19
mikehh tcurtis - one of our GSOC students, he has been working on optimizations for GSOC 22:20
Paul_the_Greek Okay, maybe I'll bring this up during Tuesday's summit. 22:22
mikehh initially his work was in the gsoc_past_optimization branch but has recently moved to github as github.com/ekiru/tree-optimization 22:24
the optimizations are being used in quite a few projects
Paul_the_Greek I've saved that link. Now I'm off for a bit of dinner. Thanks for all your help. I'll be back later.
mikehh cu
dalek rtcl-nqp: 27a67b8 | Coke++ | (2 files):
convert [namespace] to the Austin++ style of subcommand ensemble.
 22:34
purl I don't know how to convert [namespace] to the Austin++ style of subcommand ensemble..
rtcl-nqp: 4f6834c | Coke++ | README:
explain the "NQP" a bit, remove the POC. This is pretty much it now.
rtcl-nqp: d40408c | Coke++ | docs/overview.pod:
minor cleanups
rtcl-nqp: d53f3d6 | Coke++ | (3 files):
Merge branch 'master' of github.com:partcl/partcl-nqp
22:36 desertm4x joined
dalek rtcl-nqp: 11a2c80 | Coke++ | (37 files):
Avoid specifying pirop signatures if they are baked in to PCT.
 22:45
rtcl-nqp: ae8abf9 | Coke++ | (4 files):
Merge branch 'master' of github.com:partcl/partcl-nqp
 22:46
Austin That sounds like something Coke would have said. 22:47
Paul_the_Greek Hey all. 23:06
23:08 LoganLK joined
Coke (you don't need 'make clean', PTG, if you're just editing non-config files. 23:09
23:09 LoganLK joined
Coke the whole point of the makefile is so that if you change "A", "make" will rebuild only the files you need. 23:09
Austin: what does? 23:14
23:14 brianwisti joined
Coke msg kthakore there are folks checking on memory profiling, yes, but more eyes are always welcome. 23:14
purl Message for kthakore stored.
Austin Coke: something frompurl 23:16
*from purl
Coke is the person who was having trouble modifying rakudo still on? 23:18
Austin Mak? I'm not sure.
I think we got that worked out, though.
Coke++ # I got parrot to build with only a few test failures using D:\\parrot_strawberry. Which frankly ought to be pretty embarassing for parrot, but did solve my problem. 23:30
Coke msg Paul_the_Greek - why change the component on ticket # 1605? if the docs are outdated, is that not a docs issue? 23:32
purl Message for paul_the_greek stored.
Coke Austin: that's probably the least embarassing embarassing thing.
Paul_the_Greek Coke: The first round of patches were actually changes to comments and one or two bits of code, so I thought the ticket should say core. 23:33
Then when I submit the patch to the documentation, I'll change it back.
Is that the right sort of thing to do?
Coke overkill. 23:42
purl there is no such thing as overkill. there is only "not enough kill" and "enough kill" or if it's worth doing, it's worth overdoing (mythbusters) or thereifixedit.files.wordpress.com/2...257156.jpg
Coke that sort of information is much more important on tickets that are not currently being worked on. 23:43
Paul_the_Greek Okay, I won't worry about it to much. I figured it would help the folks who actually commit the patches. 23:46
too
23:47 somebody_ joined 23:51 brianwisti left
Paul_the_Greek Does the PGE make it possible to have a macro facility in a language? 23:53
plobsing Paul_the_Greek: not sure about PGE, but PCT has the ability to insert steps in arbitrary locations. 23:58
what sort of macros are we talking about? the C textish ones or the lispy tokenish ones? 23:59
23:59 Psyche^ joined
Paul_the_Greek I was thinking of text-oriented ones, but with much more power than C. 23:59
Something like a sophisticated assembler would have.

Powered by Cro and the Raku® Programming Language.

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

Thank you!