Check your feather email | spec.pugscode.org | paste: sial.org/pbot/perl6 | pugs.blogs.com | www.treehugger.com/files/th_images/paradigm.jpg
Set by audreyt on 29 August 2006.
TreyHarris markstos: sent C<when> mail to p6-l. the sub<->meth xversion mail will come later 00:09
markstos Awesome. 00:10
I'll be composing my CATCH clarification msg shortly.
First, I had to finish this post to PerlMonks that I was encouraged to make, encouraging module users to step up and help with maintenance: 00:11
www.perlmonks.org/?node_id=571832
lambdabot Title: Take Back Your Modules!
markstos TreyHarris: I just sent a refinement response to your p6l post. Thanks for starting that thread! TreyHarris++ 00:18
00:24 weinig is now known as weinig|food 00:25 rodi joined 00:27 nekokak joined
markstos I just sent my CATCH message to p6l. 00:45
00:49 weinig|food is now known as weinig 00:58 mdiep_ joined 00:59 hikozaemon joined, NamelessOne joined 01:00 khisanth_ joined, khisanth_ is now known as Khisanth 01:02 bsb joined, bcorn joined 01:06 weinig is now known as weinig|, weinig| is now known as weinig|bbl 01:08 prism joined
markstos I contributed my first "Diagnostics explanation: 01:09
rakudo.org/perl6/index.cgi?pugs_diagnostics
If you are currently stuck on "can't modify constant item", it might help you. 01:10
01:14 jon__ joined, mdiep joined
rodi markstos: on that topic: 01:16
?eval sub foo ($a is rw) { $a ~~ s/b/c/; } foo('b'); 01:17
evalbot_r13106 Error: Can't modify constant item: VStr "b"
rodi I think this is as-yet unimplimented, non?
or am I putting it in the wrong place? 01:18
markstos What's unimplemented? I think that's correct. You can't make 'b' into a rw variable.
"is rw" and "is copy" already seem to work as they should. 01:19
01:19 mdiep joined
Limbic_Region IOW - a constant string can't magiccally become mutable 01:19
rodi righto.
?eval sub foo ($a is copy) { $a ~~ s/b/c/; } foo('b');
evalbot_r13106 Bool::True
Limbic_Region if you had passed in a scalar var - it would have worked 01:20
rodi ?eval sub foo ($a is copy) { $a ~~ s/b/c/; } say foo('b');
evalbot_r13106 OUTPUT[1 ] Bool::True
rodi ?eval sub foo ($a is copy) { $a ~~ s/b/c/; } my $b="b"; say foo($b);
evalbot_r13106 OUTPUT[1 ] Bool::True
Limbic_Region change copy to rw
still will work
rodi ?eval sub foo ($a is rw) { $a ~~ s/b/c/; } my $b="b"; say foo($b);
evalbot_r13106 OUTPUT[1 ] Bool::True
markstos Limbric_Region++
rodi markstos++ 01:21
Limbic_Region (pugs hackers)++
Limbic_Region is still compiling ghc from src 01:22
1hr 42min so far
markstos ouch.
Limbic_Region hopefully subsequent builds will be much faster 01:23
markstos "make hope"
01:23 mdiep joined 01:30 frederico joined 01:35 avarab joined 01:36 weinig|bbl is now known as weinig
Limbic_Region finished - less than 2 hours 01:38
TreyHarris yee-hah, speed demon :-) 01:39
Limbic_Region once I compile Pugs with it and verify it works correctly, I will update it to verify subsequent compiles are faster 01:40
and then I will upload a bin dist for Win32
rodi Actually, it's impossible that it went that fast, because `units` doesn't understand lighthours. So there. 01:42
rodi hopes for at least a courtesy chuckle...
01:43 avarab is now known as avar
rodi should have spent a little less time on TreyHarris's 500-mile FAQ... 01:44
01:46 mdiep joined 01:47 frederico left
markstos Here's something tricky: 01:53
?eval my $str = 'a'; $str ~~ s:P5:g/(a)/{'pre'~$1~'post'}/; say $str;
evalbot_r13106 OUTPUT[prepost ] Bool::True
markstos What's trick is that while the "P5" is present, I think the {} on the right implie P6ness, which among other things, means unlike Perl 5, the captured values start at $0, not $1, hence the missing 'a' in the result. 01:54
I'm adding a mention to Differences.pod, since this certainly confuse people who think just adding :P5 is about all they need to do. 01:58
svnbot6 r13107 | markstos++ | Add a s/// example to Differences.pod 02:04
r13108 | fglock++ | Pugs-Compiler-Rule - fixed & recompiled Grammar::Rule 02:07
Limbic_Region attempting to compile pugs from source compiled GHC 02:08
svnbot6 r13109 | markstos++ | CGI.pm: escapeHTML and unescapeHTML were broken in basic ways and are now working better. 02:10
r13109 | markstos++ | two new passing tests were added for them.
Limbic_Region particle - ping 02:14
02:19 weinig is now known as weinig|zZz
statico Automatic karma with CVS commit? 02:20
markstos Subversion. But what's karma good for anyway? Can I cash it in for a stuffed animal later? 02:21
rodi err... you can run over a dogma with a karma... 02:22
statico Did I say CVS? Yikes! 02:23
markstos It might be more useful in reverse: Given enough karma, you get subversion access. :)
statico What if the commit introduces a bug?
statico trolls, harbors doubt
markstos It least you didn't say CSV, which I would have done. :) 02:24
statico: then rollback or commit a bug fix.
rodi or hit the person with your karmic bat.
markstos or spray the subversion server with pesticides 02:25
rodi About 90% of my commits introduce at least one bug,
since they are usually failing tests. Though I did, admittedly, fix the sprintf one today :-D
rodi preens...
s/are/consist of/ 02:27
02:28 shasbot joined
audreyt wow, I'm going to be late at $work. 02:29
audreyt spent 30+ minutes rofl'ing and generally disabled by TreyHarris's 500mi story 02:30
markstos A more mundane pugs blog post today: pugs.blogs.com/pugs/2006/09/cgiapp_...rl__4.html 02:34
rodi well, as one ore more of my babies are likely to wake me in four or five hours, I'm going to go to bed now. Until next time, fare thee well, my e-buddies. 02:36
02:37 markstos left
audreyt sleep well, rodi 02:37
clkao audreyt: yo. jifty. need a way to have dynamic arguments based on the current state of action. defer doesn't get the action itself. 02:39
clkao . o O ( CALLER::$?SELF ) 02:40
audreyt do what we do in dispatcher? 02:44
i.e. have a package global that stands for the current action.
seems to me cleaner (far cleaner) than walking the call stack and pull $_[0] out. 02:45
clkao ah. cunning. where are we doing that?
audreyt local our $Action = ... 02:46
grep for $Dispatcher in Dispatcher.pm
02:51 shasbot left 02:54 Limbic_Region joined
Limbic_Region particle ping 02:54
pugs built successfully btw 02:55
oh, audreyt - you're about
great news
I am compiling ghc from source on Win32
hrm - can't tell if that late for $work comment meant she was in transit or what 02:56
02:59 Termy joined
Limbic_Region particle - if you backlog, feather.perl6.nl/~limbic/ 03:07
audreyt ping feather.perl6.nl/~limbic/ # see above for explanation 03:08
ok, time for bed
03:13 mako132_ joined 03:14 MacVince joined
MacVince Quick question. I just saw a post on the perl6 planet blog about CGI.pm. The guys said that Perl 5's $1 becomes $0 in Perl 6. So what's $0 in Perl 6? 03:15
Meaning, how to I get the name of the program
dolmans MacVince: $*PROGRAM_NAME 03:27
MacVince Cool 03:28
dolmans MacVince: do you use Pugs? in t/magicals there are some magicals variable examples. 03:30
04:49 drbean joined 05:03 kanru joined 05:24 iblechbot joined 05:34 hikozaem_ joined
nothingmuch 9morning 05:42
gaal 8:60morning 05:48
(I see your typo and raise you a leap minute!)
05:55 MacVince left
nothingmuch gaal: ? 06:08
oh
me doesn't parse well at these hours 06:09
06:10 mugwump joined
gaal if I ever end up writing a parser engine, I think I'll call it caffeine, since it's my experience that parsing is often imeded by the lack thereof 06:10
*impeded 06:11
typing, however... I don't know of a substance that improves it. 06:12
mugwump how about ibogaine? it's a strange substance used by shamans in parts of africa to cleanse the brain during rites of passage ceremonies 06:17
gaal that sounds dangerously like iocaine
06:30 hermax_ joined 06:31 christopher joined
svnbot6 r13110 | malon++ | smokeserv-client.pl - 06:31
r13110 | malon++ | * Fail if .yml is not provided
r13110 | malon++ | * Check that the configurations used for the .yml and .html correspond
r13110 | malon++ | * Accept .yml and .html arguments in either order
r13110 | malon++ | run-smoke.pl - use a .yml name like the .html, not constantly tests.yml
r13110 | malon++ | Makefile.PL - make upload-smoke depend on smoke.yml instead of tests.yml
christopher If you have problems submitting your .yml smoke, let me know (by IRC or email). 06:32
I was surprised that the r13105 win32 smoke didn't send .yml properly. 06:33
06:54 mj41_ joined, NamelessOne joined
gaal audreyt: ping 07:01
need some ghc typing help 07:02
07:03 NamelessOne joined
gaal I *think* this is a ghc bug, in fact :( 07:08
araujo hi 07:09
gaal ho
araujo gaal, which one?
gaal which bug? which ghc?
ghc is having problems unifying types of an implicit parameter and a typeclass-constarined alias 07:10
*constrained
ghc-6.5.20060818
07:10 iblechbot joined
araujo i _highly_ recommend you to update gaal 07:12
gaal it's not that old... but yeah, good idea 07:13
araujo gaal, they have recently fixed some bugs 07:14
araujo installed yesterday snapshot and now will update again 07:15
gaal hmm, okay, but if indeed this is one of them then this feature won't compile on 6.4.x :(
araujo well, they expect to release new version soon afaik 07:18
gaal yeah.
I just hope it'll come with good bindists for all platforms, so that we can update our dependency immediately. 07:19
(I already have another change in a branch waiting on the move...) 07:20
well, that didn't work. scping patch... 07:35
audreyt: could you give perlcabal.org/~gaal/tmp/priggy.0.patch a gander? 07:37
the problem occurs in Prim, binding the new implicit parameteric pretty-printer. 07:38
(background w/motivation: .perl should not ultimately show things like 'VPure (NInt 42)', but rather just '42'. But for development we really want some prim to dump the guts with priggy attention to detail.) 07:42
07:47 kane-xs joined 07:49 bsb left 07:53 rindolf joined
rindolf Hi all. 07:53
07:57 buu joined
mugwump anyone got an opbot I can use? 08:00
nothingmuch mugwump: BingOS on irc.perl.org runs flexo/bender 08:13
maybe you could ask him for the code 08:14
08:27 drrho joined, ruoso joined 08:37 ruz joined
rindolf Hi nothingmuch 08:37
nothingmuch hi ho
rindolf nothingmuch: what's up? 08:41
nothingmuch meta models 08:42
08:44 buu joined
rindolf nothingmuch: meta models of what? 08:49
nothingmuch of meta models
meta meta models ;-)
of perl 6 / moose object space, actually
rindolf nothingmuch: I see. 08:50
08:51 jferrero joined
rindolf nothingmuch: I'm working on an XSLT transformation. 08:51
nothingmuch: I think I finally grokked XSLT, after I read the W3Schools and Zvon tutorials. 08:52
08:55 avarab joined 09:17 iblechbot joined 09:22 polettix joined 09:44 foo\ joined 09:49 elmex joined 09:55 ludan joined 10:02 NamelessOne left 10:06 bradleym joined 10:38 prefiks joined 11:17 chris2 joined 11:32 Termy joined 11:44 Alias_ joined
Alias_ nice paradigm thingy 11:44
11:53 mako132_ joined 11:57 marmic joined
rindolf Hi Alias_ 12:03
12:03 ruz joined, Limbic_Region joined
Alias_ hiya 12:04
How goes the Error.pm saga
rindolf Alias_: well LeoNerd took over most of the development. 12:05
Alias_ (It's pass'ing on Strawberry now btw)
ah
rindolf Alias_: GBARR (IIRC) made me the primary maintainer. 12:06
Alias_: and LeoNerd a co-maintainer.
Alias_: do you still live in .au ? 12:07
Alias_ All that bubbled up to me was some wailing and nashing of teeth after it got broke'd
Yeah, in Sydney
rindolf Alias_: OK.
Alias_: I fixed two bugs yesterday and today: one in TAPx::Parser and one in XML::SemanticDiff. 12:08
12:15 weinig|zZz is now known as weinig
Limbic_Region Juerd ping 12:16
rindolf Alias_: what's up with you? How is the Perl refactoring tool progressing? 12:24
Alias_ I got distracted by making a Win32 Perl distribution
12:24 xinming joined
Alias_ So I could run it on my laptop 12:24
12:24 christopher joined
Limbic_Region salutations rindolf and Alias_ 12:25
Alias_ Now that the main group of important modules are all mostly green on Win32 now (Wx and so on) I'm trying to find the time to keep going on the editor part
But herbert can be hard to work with (shifting decisions)
Xeper -> Kepher -> Kefra etc
But as the Win32 Perl stuff gets stronger and stronger, at least I'm not being worried by the base level stuff giving me problems 12:26
rindolf Hi Limbic_Region 12:30
Alias_: I've been tutoring someone in Perl using Windows this week.
Alias_: I found a way to segfault XML::LibXML. 12:31
If I specified the ISO-8859-8 encoding for my XML.
12:31 buetow joined
Alias_ On Windows? Doesn't sound hard 12:32
How did you get it installed on Win32?
(that's one of our ones with the external modules issues)
eep, afk now 12:33
(talk to you later)
rindolf Alias_: I installed XML::LibXML by configuring the uwinnipeg ppm repository and then hitting ppm install XML::SAX and ppm install XML::LibXML. 12:34
13:00 agentzh joined
agentzh christopher: the only SYN link on the smoker sever seems to have some problems. :( 13:01
13:01 crem_ joined
agentzh the smoke results are very odd. 13:01
christopher agentzh: I think iblech stopped his smoke after 13 tests, so most things say "0. checks, 0 X's." 13:03
agentzh okay 13:04
is the smoke server using the latest synopses?
what is /var/www/iblech/stuff/pugs-smokes/spec?
christopher I think so. I'm not sure how often the refresh-synopsis script is being run. 13:05
That's a directory that should be getting updated with the synopses.
agentzh i think you should not specify the --syn-dir option your self.
or i even should not offer that option.
christopher It uses the same update script found in docs/Perl6/Spec in the pugs tree.
But I don't think the entire pugs tree is checked out on the Smoke server. 13:06
agentzh christopher: why not use the default setting?
christopher The important part is that something's auto-updating the specs, wherever they are.
agentzh or the syn rev number won't be set in the HTMLs
christopher: what is updating that? a cron? 13:07
christopher Yeah, the rev number is a problem. For some reason, version_h.pl doesn't seem to work in isolation.
agentzh christopher: smartlinks.pl automatically update the synopses it uses everytime it runs (unless you specify the --fast option)
christopher Yes, it's meant to be updated by refresh-smoked-synopses.pl, running under cron.
agentzh *updates
so you don't have to do that yourself. 13:08
christopher yes, by running the "update" script which we call every time.
We do it ourselves because refresh-smoked-synopses will update all the smoked synopses at once. When done in a batch, it's faster to just call the "update" script once.
agentzh christopher: i don't understand. 13:09
you're keeping various revisions of synopses on the smoke server? 13:10
christopher agentzh: When --syn-dir isn't in the pugs tree, you need to run "update" yourself, or smartlinks won't do it for you.
No, we only keep the latest (unsmoked) synopses on the server 13:11
But we produce a different set of smoked synopses for each smoke report.
agentzh so i suggest you use the synopses *in* the pugs tree. :)
that is *strongly* recommended.
christopher When the synopses get updated, the refresh script reruns smartlinks on all the previous smoke reports, to give them the latest synopsis.
agentzh christopher: on all the previous smoke reports? is that necessary? 13:12
13:12 cjeris joined
christopher agentzh: but I don't think there is a full pugs tree on the smoke server. Would you prefer we checkout pugs/util? Will that be enough for all your version-watching scripts to work? 13:12
agentzh christopher: you don't need to check out the whole pugs tree. 13:13
christopher It's necessary if you always want to be looking at the latest synopses. You obviously can't look at the latest tests on an old smoke, but you should always want to compare them to the latest spec.
agentzh christopher: okay, i can understand that. 13:14
checking out pugs/util is not enough at all. 13:15
[particle] hrmm, i'd expect you'd want to look at the spec that's at the same pugs rev as the tests
agentzh particle: yeah, i'm having exactly the same feeling. 13:16
christopher: is checking out a whole pugs tree *that* bad?
you always have cron, you know.
christopher particle: I envision people mainly want to see how close they are to the current spec, by whatever means are available to measure it. Meeting an old spec would be like a report-card sort of satisfaction. 13:17
Limbic_Region [particle] - did you get my /msg
woot
[particle] lr: yes, haven't acted on it yet, as i've only been awake for 17mins
:)
Limbic_Region unfortunately, feather is unlikely a good place to host the binary dists as I have been downloading the 54 MB for over an hour now 13:18
christopher agentzh: You can suggest it to iblech. I don't really mind. Do you understand why we need the whole tree, though? (On my machine, it takes 30 seconds to see what files have been updated, from top level, when doing "svn status," so I can see why some people wouldn't want the whole tree.)
Limbic_Region assumes the issue is on feather's end
[particle] christopher: why not have a local mirror of pugs, and use svk to pull the proper rev?
then you have *all* revs locally 13:19
christopher particle: Is this a way svk is faster than svn? It may be worth installing svk on the smoke server, then. (It's not under my administration.)
agentzh: Is the version information you print a synopses revision number, a pugs revision number, or both? 13:20
agentzh christopher: fyi, there's a perl script on feather which updates the whole pugs tree every 5 min or so. :)
[particle] i think having a local mirror is faster than net traffic
agentzh christopher: both.
christopher: look at here: svn.berlios.de/svnroot/repos/unisim...n/S02.html 13:21
[particle] agentzh: i've started modifying smartlinks for parrot
lambdabot Title: S02, tinyurl.com/lcojf
agentzh particle: woooot!
[particle] imo --syn-dir is important to me
agentzh particle: why?
[particle] in fact, i want to make it --syn-path 13:22
because i have two sets of specs to check
PDD* and S*
agentzh ah, gotcha. :)
[particle] S* are not (currently) checked into parrot repo
christopher agentzh: I think the current smartlinks assumes that the revision of the pugs tree in which it lives, is the revision of pugs that should be printed. We wouldn't want to do that on the smoke server, because those revision numbers would keep updating, and the smoked pugs revision should be frozen.
[particle] christopher++ # if that's the case, it needs fixing 13:23
agentzh christopher: then we need a better way to detect rev number.
christopher agentzh: so I think smartlinks.pl should read its pugs revision from the .yml, if present, not from its ambient pugs tree.
[particle] agentzh: it should be in the smoke report
13:23 mdiep_ joined
agentzh right! 13:24
when .yml is provided, get pugs rev number from there.
[particle] Limbic_Region: 6kps -- expected 2:21:00 remaining :(
agentzh christopher: okay, i'll add that to smartlinks.pl tonight. 13:25
christopher: but how about the rev number of synopses? 13:26
christopher agentzh++
Ah, you're getting the synopses by http (and so are we)
agentzh christopher: i still think we should checkout at least pugs/docs/Perl6/Syn
christopher If we did a real svn checkout of the synopses we could have the synopsis revision number. 13:27
agentzh christopher: could you check out pugs/docs/Perl6/Syn?
Limbic_Region I am getting 9
tonight, I will put it on my own website 13:28
christopher agentzh: are you getting synopses by http just because you worry people who get the pugs Release versions may not have svn installed on their machines?
Limbic_Region and if that still is no good - I will try perlmonk.org
agentzh christopher: aye.
Limbic_Region unless you have somewhere you would want me to upload it directly?
agentzh christopher: that's something about "laziness".
i just want to resue the pugs mechanism as much as possible.
yes, http is slower. 13:29
but that may not be a problem. 13:30
christopher agentzh: Can the .svn subdirectory (with metadata) of some place that does svn checkouts, be accessed by http?
We just need to download our synopses from somewhere that also makes the corresponding revision number easily readable.
agentzh christopher: from feather? i dunno. 13:31
rindolf Alias_: ping 13:32
13:32 rodi joined
christopher agentzh: Yes, feather would be natural, if it's auto-downloading the synopses anyway. 13:32
agentzh christopher: it's updating synopses every 1 hour.
...err, by cron. 13:33
[particle] Limbic_Region: i don't, and i don't mind the wait. hopefully i won't need to get every daily snapshot of ghc to build pugs :)
Limbic_Region [particle] - oh, you definately won't
christopher agentzh: If there were feather.perl6.nl/syn/.svn, then you could just read the .svn/entries file the same way pugs does, and figure out what revision it was at. 13:34
Limbic_Region [particle] - you ever used Parallel::ForkManger
[particle] Limbic_Region: yes, about 6 years ago i used it a lot :)
agentzh christopher: please please don't use svn to check out synopses...that would make my life particularly harder... :)
christopher: i have a suggestion. 13:35
christopher agentzh: No, I'm not suggesting that on the client-side. I'm only suggesting that _feather_ uses svn to check out synopses, instead of using wget.
Limbic_Region [particle] - I ask because I am updating util/yaml_harness.pl to use Parallel::ForkManager if available (so Win32 can do smoke tests in parallel) and only fall back to the homegrown version in the event it isn't present
I was wondering if you were interested in helping 13:36
agentzh christopher: why not using LWP?
christopher Then the client gets svn synopses from feather (instead of from the perl repository directly) and gets their associated revision number.
[particle] i think i can help with that
Limbic_Region fwiw, P::FM isn't foreign to me - I am just more likely to be motivated if I have someone suggesting ideas and helping 13:37
I'm terribly lazy otherwise
[particle] i have the 0819 ghc build now, but haven't built pugs with it yet
i'll do that now (no sense waiting for 0906)
do you have work-in-progress checked in?
Limbic_Region no, but I can nopaste it
[particle] thanks to juerd++ we can use *obby on feather
and edit the file together 13:38
agentzh christopher: i don't think the client should get svn synopses from feather. feather updates syn every 1 hour, which is not very frequent.
[particle] agentzh: do you really think the synopses often change more than once per hour? 13:39
christopher agentzh: The smoke server won't update synopses more than once an hour, probably.
agentzh: How is the synopsis revision computed currently by smartlinks.pl?
agentzh particle: i think the user always wants the latest, even psychologically.
Limbic_Region [particle] - what is the gobby URL 13:40
agentzh christopher: smartlinks.pl gets the syn rev number by invoking util/version_h.pl
[particle] hrmm, i wonder if an rss feed could be set up so folks can watch for syn updates and act on changes
lr: just know sobby debian pkg was installed last night
agentzh christopher: oh, sorry...
Limbic_Region and btw - I am not just shoe-horning in P::FM, I am pretty much re-writing it
christopher agentzh: which does...? (I understand the pugs svn magic but not the synopsis magic)
[particle] i haven't logged into feather in a while 13:41
oh, rewriting p::fm! this should be... fun :)
Limbic_Region ok - no worries, I will nopaste for now and we will bug Juerd when he shows up
no, not p::fm - yaml_harness
[particle] ah
Limbic_Region lines like @ARGV = sort map glob, "t/*/*.t", "t/*/*/*.t", "ext/*/t/*.t" if ! @ARGV; # bother me
agentzh christopher: sorry, by reading docs/Perl6/Spec/.spec-revision
that file is generated by the "update" script. 13:42
pasteling "Limbic_Region" at 129.33.119.12 pasted "[particle] - here is my work in progress on util/yaml_harness.pl" (293 lines, 8.6K) at sial.org/pbot/19541
13:42 weinig is now known as weinig|bbl
[particle] builds pugs with ghc-6.5.20060819 and AS perl-5.8.8b817 13:43
christopher agentzh: Good, it looks like it comes in the http response. So I think pugs tree checkout is unnecessary.
agentzh christopher: could you check out docs/Perl6/Spec? that is required.
Limbic_Region [particle] - my mental hurdle at the moment is figuring out how to modularize P::FM or the homegrown solution as plug-ins so either will work
christopher agentzh: by what? I think only that "update" script is required. 13:44
Limbic_Region I was hoping to overcome that hurdle today but I had baby duty last night - no sleep
13:44 NamelessOne joined
rodi Limbic_Region: me too!! 13:44
agentzh christopher: the *path* is also required.
rodi err... not the hurdle, but the baby ;)
agentzh christopher: so please make sure the update script is in the right path. that is very very important. 13:45
Limbic_Region [particle] - additionally, [audreyt] said it was ok to assume perl 5.8 when rewriting so no need to keep 2 arg open, local *fh instead of lexical, etc
[particle] Limbic_Region: this is the first time i've seen this code... is the homegrown solution in this file?
agentzh or we'll need to modify smartlinks.pl.
[particle] i thought 3 arg open was a 5.6.1 thing
Limbic_Region [particle] - yes, my $s = __PACKAGE__->new; $s->run; $s->emit;
[particle] k 13:46
christopher agentzh: I think you should modify smartlinks.pl to look for $syn_rev in a file called $syn_dir/.spec-revision (just move that part outside of your "dir eq dir" block)
Limbic_Region [particle] - my point is there is no need to preserve compatability < 5.8
[particle] btw i'd put all subs below exit; but that's just my style
agentzh christopher: i second that. will do tonight. :)
[particle] Limbic_Region: easy way, 'use 5.008;'
agentzh christopher++
Limbic_Region [particle] - that doesn't actually use those features though ;-) 13:47
[particle] have perl worry about compat, so you don't have to
agentzh christopher: btw, thank you for your excellent work!
Limbic_Region [particle] - I was arranging the subs by the order I had "touched" them versus ones I still needed to touch
with the intention of re-arranging once they were all given a once over
[particle] gotcha... work-in-progress
those above have been touched?
13:48 avarab is now known as avar
[particle] or below? 13:48
Limbic_Region yes
christopher agentzh++ (smartlinks is so cool it inspires lots of add-on hacks)
Limbic_Region below is untouched
agentzh christopher: :)
[particle] christopher++ # i totally agree
agentzh (bbiab)
Limbic_Region though fix_config() is unfinished - the empty else {} block and the if above it are where the P::FM and homegrown plugins need to go 13:49
s/need to go/should go I think/
[particle] yeah
and you only need P::FM if it's win32, right? 13:50
Limbic_Region no
[particle] or are you planning to replace homegrown with P::FM entirely
Limbic_Region my thinking is we should use P::FM regardless of platform if it is available
13:50 jferrero joined
[particle] ok 13:50
Limbic_Region if it isn't - then homegrown and Win32 is SOL
If the file was self-contained I don't think I would be having the brain cramp but SUPER methods are being called 13:51
[particle] well, if the eval fails, just set $Config{"concurrent"} to zero if platform is Win32
Limbic_Region and I don't know the test harness stuff all that well
no to 1
concurrent isn't boolean
[particle] oh
Limbic_Region it is the number of tests to run concurrently
[particle] :)
that's not yet obvious 13:52
Limbic_Region was confused on that matter too until I started digging into the homegrown code
[particle] the usage says it, i just didn't read it
btw typo in usage
Paralle::...
13:54 kanru joined
Limbic_Region maybe I am just being overly idealistic with converting to plug-ins. It is relatively simple to just complete two completely different branches and have code duplication 13:54
13:56 weinig joined, weinig is now known as weinig|bbl
Limbic_Region [particle] oh, I also wanted to use a proper temp file module if available too as the current homegrown solution doesn't always clean up all the files 13:57
I think it is an issue of no sig handlers but I can't be sure
agentzh christopher: please don't forget to set the --fast option while invoking smartlinks.pl on the smoke server in case that you invoke the "update" script yourself.
[particle] yeah, see that 13:58
christopher agentzh: will do!
[particle] (temp file) *yuck*
agentzh christopher++
Limbic_Region and come to think of it - since < 5.8 compatability is being dropped - the shuffle routine can go away completely and be replaced with use List::Util 'shuffle'; since it has been core since 5.7 13:59
Limbic_Region just likes messing with the symbol tabls
table even
[particle] :)
isn't file::temp core?
Limbic_Region not sure - check Module::CoreList
[particle] doing that
Limbic_Region too - slowly 14:00
yes, since 5.6
not sure why it wasn't used originally then
[particle] easy win
false laziness
14:01 mako132_ joined
Limbic_Region oooh, I am at 95% of my download 14:01
[particle] i'm at 22% and 4.7kps
Limbic_Region so [particle] - when you get bored of playing with it and/or need to work on something else - just /msg me back the nopaste
[particle] geez, that's 4800baud
Limbic_Region or better - email me 14:02
agentzh christopher: please test r13111. (i've tested on my local machine.) 14:03
svnbot6 r13111 | agentz++ | [util/smartlinks.pl]
r13111 | agentz++ | - checked the existence of the "update" script in user-
r13111 | agentz++ | specified syn path instead of mechanically comparing
r13111 | agentz++ | the path with pugs/docs/Perl6/Spec, requested by
r13111 | agentz++ | christopher++
agentzh err...step #2, read pugs rev number from .yml (if any)... 14:04
14:05 onsen joined
svnbot6 r13112 | malon++ | smokeserv - call smartlinks.pl with --fast (noticed by agentzh++) 14:05
14:10 lisppaste3 joined
agentzh step #2 is done. 14:16
svnbot6 r13113 | agentz++ | [util/smartlinks.pl] 14:17
r13113 | agentz++ | - reads pugs rev number from the smoke data (.yml) if
r13113 | agentz++ | --test-res is specified. this is requested by
r13113 | agentz++ | christopher++ and particle++.
christopher agentzh++ (thanks!)
14:17 prefiks left
agentzh christopher: to be honest, i'm *so* eager to see the first set of smoked synopses on the smoke server. :P 14:18
christopher agentzh: I'd submit one, but "make smoke" stalls for me. 14:19
agentzh hehe.
i've never run smoke on my machine. :)
christopher On t/examples/all_parse.t, it hangs for hours. If I interrupt, I get no smoke at all.
agentzh ...just in fear of hanging. 14:20
*nod*
christopher And if I run "pugs t/examples/all_prase.t," I get "Illegal instruction."
Should yaml_harness.pl time out after a fixed amount of time on any test? 14:21
agentzh no clue...i'm not familiar with the smoke facility. 14:22
14:22 amnesiac joined
christopher I'd say "yes," but all_parse.t is a special test that runs many examples, so maybe its timeout threshold should be bigger. 14:22
agentzh i hope it does.
christopher AFAICS it doesn't time out at all.
And this is misleading, when the smoke reports can't even be sent in the first place to report the failure. 14:23
[particle] perhaps ipc::run could be helpful
agentzh if there's timed out mechanism, we can even run autosmoke on feather.
14:24 rindolf left
agentzh i have no idea how audreyt works around hanging while smoking. 14:24
christopher particle: Is that more portable than "wait", "alarm", and friends? 14:25
[particle] probably not
but it offers a lot
it is perhaps overkill 14:26
christopher I'm not familiar with it.
agentzh IRC::Run is not very friendly on Win32.
[particle] think of it as IPC::Open3 that works
agentzh: i use it in production
agentzh for windows, IPC::Run3 is better, though not very featureful.
[particle] it's not working on win98, but i have no problems in win2003 14:27
agentzh particle: i once used IPC::Run to rewrite the gnu make test suite (on win2000), but it suffered a lot.
kane-xs ipc:: on win32 is painful
basically taht which works on every other OS fails on win32
agentzh kane-xs: IPC::Run3 is a notable exception. :) 14:28
kane-xs it lacks a lot of features
that's a good way to make things work more
agentzh kane-xs: yes.
windows-- 14:29
[particle] (perl threads)-- 14:30
agentzh heh
14:32 frederico joined 14:33 frederico left 14:39 dakkar joined 14:42 vel joined 14:47 elmex joined
svnbot6 r13114 | agentz++ | [t/closure_traits/in_loop.t] 14:51
r13114 | agentz++ | - updated the tests to reflect recent syn changes.
r13114 | agentz++ | Status:
r13114 | agentz++ | all tests are failing.
r13115 | audreyt++ | * Remove the now-obsolete support for precompiled-in-haskell
r13115 | audreyt++ | prelude, since it's redundant with YAML-based prelude support,
r13115 | audreyt++ | and (with GHC 6.6) the speed difference is indistinguishable.
agentzh when should LAST {} be triggered? will return or leave bapasses LAST {}? what about the case when there's no explicit "last"?
*bypass
14:52 dakkar joined
audreyt no explicit last should trigger LAST all the same 14:54
agentzh auderyt, TimToady: what are the exact running circumstances for LAST {}? only if there's a "last" or the end of the loop is reached normally? just like NEXT {}?
audreyt: will return/leave/goto bypass LAST {}? 14:55
audreyt I think all three will 14:56
agentzh oh, S04 is still quite hand-waving. :/ 14:57
audreyt indeed.
agentzh maybe it's time for me to start another p6l thread. :)
audreyt maybe it's time to draw some pictures :) 14:58
but p6l thread is a good idea
agentzh okay :)
audreyt :) 14:59
agentzh often gets blocked by the handwaving synopses while writing tests for pugs. 15:01
audreyt you need nonblocking IO 15:04
i.e. write the test the way you think it ought to work 15:05
and then ask #perl6 and p6l :)
agentzh audreyt: good idea. :D 15:08
audreyt iblech used to do that all the time :) it's a very good approach 15:10
15:10 weinig|bbl is now known as weinig
agentzh iblech++ 15:11
i can imagine the painful early days of pugs... :) 15:12
audreyt it's actually a lot of fun :)
significantly more frustration than fun at the beginning, though
it's getting much better :)
agentzh nod nod 15:13
audreyt looks at gaal's priggy.0.patch
Juerd Limbic_Region: Pong
agentzh so i was lurking from pugs day #1 to this july.
Juerd Limbic_Region: feather should be good for binary dists. If it's slow, please mail me your IP so I can traceroute. 15:14
agentzh fyi, feather was pretty fast while i was downloading a 2 GB file from gaal's home. 15:15
[particle] juerd: can i msg, or is email better? 15:16
audreyt gaal: I _think_ implicit parameters wants to be monomorphic
but I also think it's better to use a ReaderT.
15:16 frederico joined
[particle] juerd: sent email 15:18
Limbic_Region Juerd - both I and particle are extremely slow 15:22
behind a nat'd firewall so will take a sec to get my IP 15:24
audreyt gaal: however, if you do want to be polymorphic, the one line fix is: 15:25
type PrettyPrinter = (forall a. Pretty a => a -> String)
the explicit foralls will guard the "a" from escaping toward universal quantification. 15:26
er, nvm, it needs to be a newtype.
svnbot6 r13116 | agentz++ | [t/closure_traits/enter_leave.t] 15:30
r13116 | agentz++ | - added this new test file to test ENTER and LEAVE blocks
r13116 | agentz++ | Status:
r13116 | agentz++ | all tests are failing now.
audreyt gaal: I'll commit your .guts patch 15:31
gaal: as you can see, the fix is simply to invent a newtype to put existential quantified functions into it. 15:32
Limbic_Region Juerd - email sent 15:33
svnbot6 r13117 | audreyt++ | * Commit the .guts patch from gaal++ to ease debugging.
r13117 | audreyt++ | Not meant to be tested or made into public interface!
Juerd OrgName: Computer Sciences Corporation 15:34
Sound familiar?
Limbic_Region Juerd - my IP you mean? I can't really discuss it
but I verified it 3 different ways so I am sure that is it
Juerd Nah, this is jerry's last hop. 15:35
Limbic_Region: Is leeching from other dutch sites fast?
For example speedtest.zylon.net/100mb.bin
Or proserve.nl/100mb.bin
Both traceroutes end in too-firewalled-to-be-useful networks 15:36
But until reaching those networks, speeds are okay.
15:36 spinclad joined
Limbic_Region let me check 15:36
15:36 NamelessOne left 15:37 Revision17 joined
Juerd I can leech at 80 Mb/s from feather, from another datacentre. 15:37
Limbic_Region Juerd - it is slow but not nearly as slow as feather - probably around 20kbs 15:38
though it is steadily climbing
the 8kbs for feather was average over the entire 54mb I downloaded 15:39
having downloaded 4mb or so from zylon.net - I am at around 30kbs
Juerd Limbic_Region: Sorry, it is a local thing for you then. 15:41
svnbot6 r13118 | agentz++ | [t/closure_traits/enter_leave.t]
r13118 | agentz++ | - added many more tests for explicit return and leave.
Limbic_Region smiles as he is about to build Pugs from ghc 6.6 rc (2006-09-06 built directly from source)
Juerd Perhaps you're both indirectly connected to the same backend.
Limbic_Region well, that must be most of the US then
audreyt ?eval vv 1 15:42
15:42 evalbot_r13106 is now known as evalbot_r13117
evalbot_r13117 CCall "perl" CaptMeth {c_invocant = VPure (IFinite 1), c_feeds = [MkFeed {f_positionals = [], f_nameds = {}}]} 15:42
Limbic_Region heads out for lunch but thanks Juerd for investigating first
nothingmuch mooooooooo 15:44
nothingmuch takes a nap
audreyt: want to see an AG runtime with no prettiness?
i'd like comments if you have the cylces, but very low prio
rodi In reading S06, I came across the syntax `proto sub foo {...}`, which appears not to be valid in pugs... in t/blocks/multi_sub.t, there is alternative syntax: `proto mmd`, which works- is one of these now invalid? Should I add a test for the former? 15:47
(mmd being the sub name in the latter) 15:48
audreyt nothingmuch: tomorrow definitely. mail me some urls and/or hivemind me?
nothingmuch naaah
it'll be something/somewhere else by then
agentzh audreyt: could you fix the mysterious parsefail in t/closure_traits/keep_undo.t for me? it's very annoying. 15:49
it's related to UNDO, i guess. 15:50
audreyt rodi: please grep for a test 15:51
pasteling "agentzh" at 210.22.200.67 pasted "Win32 build failure" (11 lines, 461B) at sial.org/pbot/19543
audreyt and if not there then write one
rodi audreyt: already did, which is how I found t/blocks/multi_sub.t 15:52
cool.
integral audreyt: theory's asking for someone to remove his ban in here 15:53
agentzh audreyt: please look at the build failure paste. 15:54
not sure if i should run realclean.
15:55 ChanServ sets mode: -b justatheory!n=justathe@dsl093-038-250.pdx1.dsl.speakeasy.net!##unavailable
audreyt did so. 15:55
agentzh: don't; fixed
agentzh oops! too late...
;-)
audreyt: i was caught by another problem while running `perl Makefile.PL': 15:56
svnbot6 r13119 | audreyt++ | * Parse for "proto sub foo {...}" as noticed by rodi++. 15:57
r13120 | audreyt++ | * (minor) Further refine the prelude loading communication via env vars.
agentzh *** ActivePerl 5.8.7 and below does not support Perl 5 embedding.
audreyt: must i upgrade my activeperl?
rodi LOL
I haven't even finished my tests, svnbot6! 15:58
audreyt agentzh: it's recommended, yeah
agentzh okay, that will be a long night then. :)
15:59 hcchien joined
svnbot6 r13121 | audreyt++ | * Attempt to unbreak 6.4.1. 15:59
16:00 justatheory joined
justatheory Sorry for all the noise from my stupid PowerBook falling asleep and waking all night the other night. 16:00
rodi Am I to infer from L<S06/"Routine modifiers"> that proto is also allowed as a modifier to any(<sub method submethod regex token rule macro>) ?
audreyt that would be a correct inference. 16:01
rodi coolio.
agentzh rodi: or even L<S06/Routine modifiers/Prototypes> :) 16:02
rodi :)
agentzh i didn't upgrade to 5.8.8 when it was released due to the Module::Install problem. 16:03
16:04 weinig is now known as weinig|bbl
agentzh i think now it's vanished already. 16:04
[particle] agentzh: i just upgraded to AS 5.8.8b817
is there something i should look for?
agentzh particle: good to learn. :)
[particle] btw my pugs build failed :( 16:05
agentzh particle: some old modules using old Module::Install couldn't compile at ActivePerl 5.8.8.
[particle] ah 16:06
well, if there's a specific one, i can try it
pasteling "[particle]" at 144.81.84.199 pasted "failing win32 pugs build" (213 lines, 10.7K) at sial.org/pbot/19544
[particle] looks like pugs is getting confused with gcc / msvc
agentzh particle: i didn't remember. sorry.
[particle] np 16:07
16:07 penk joined 16:08 bernhard joined
[particle] er, wait. 16:08
agentzh particle: it's not failing.
[particle] i think pugs built, those are not errors, just warnings
duh
agentzh lol
[particle] how long for make test?
agentzh 8 min for audreyt 16:09
[particle] k. started at :08, i'll check it in a few mins
agentzh thinks particle has a shinny configuration for pugs...activeperl 5.8.8, ghc 6.5... 16:10
[particle] yep! 16:11
agentzh *shining
:)
[particle] of course, i need to install half of cpan now for my other progs
justatheory Ooh, audreyt, you're here! 16:12
svnbot6 r13122 | rodi++ | Sanity checks for use of proto modifier.
16:12 Psyche^ joined
agentzh particle: that's the sad part. :) 16:12
[particle] why is 'nmake test' after 'nmake fast' building pugs ? 16:13
svnbot6 r13123 | agentz++ | [t/closure_traits/keep_undo.t]
r13123 | agentz++ | - more tests for multiple KEEP/UNDO blocks
agentzh particle: you mean the judy compilation?
[particle] i'm not sure 16:14
audreyt [particle]: it always relinks pugs
which should be fast
[particle] this is not fast, 6min so far
agentzh i'm planning to offer a Makefile for judy, which can save much time. 16:15
pasteling "[particle]" at 144.81.84.199 pasted "what does this mean?" (444 lines, 23.3K) at sial.org/pbot/19545
agentzh oh, it "nmake" normally 16:16
maybe the "test" target should depend on "fast"?
the default target is very time consuming.
16:17 dada joined
audreyt once we depend on 6.6, yes, sure :) 16:17
dada hello world
agentzh hello dada
[particle] so, i can't test an unoptimized pugs? or just need to run prove instead of 'make test'
dada quick question: is "has $.attr = 100;" supported yet? (that is, assigning a default value to an attribute) 16:18
agentzh just can't remember the last time he ran 'nmake test'.
dakkar dada: try and see what happens? ;-)
dada another quick question: it seems that all attributes are rw by default, and they shouldn't. am I right?
dakkar: unexpected "=" happens :-)
audreyt dada: it is not; workaround is to have a submethod BUILD ($.attr = 100)
dada: you are correct; the object model is currently nonexisting; a new one will be swapped in right after reelase. 16:19
dada I see
maybe throwing in a test or two for such things would help, however
audreyt currently the object model is same as perl5's, basically
tagged hash and such
it will help, but I think those two are tested under t/oo/. if not, please commit
(do you have a commit bit?) 16:20
dada the tests mentioned in the synopsis don't cover this
I had it, I don't know if it's expired
audreyt then it's missing smartlinks
commit bits, once accepted, does not expire
agentzh dada: we don't have sufficient smartlinks. ;-)
justatheory audreyt: Did you see my post about evil operator overloading on p5p?
audreyt please by all means populate t/oo/ with smartlinks
dada ok, I will hunt down the tests then :-)
audreyt very important for 6.28.x
agentzh asks audreyt when the next release will be. 16:21
pugs release == test suite triage 16:22
16:24 Psyche^ is now known as Patterner
audreyt agentzh: if it were not the macbook crash, it should be now 16:24
[particle] $ sobby 16:25
Address family not supported
agentzh ohhhhhh
audreyt with the crash, I'll releng over the next few days
agentzh macbook-- macbook-- macbook--
audreyt and aim for a release next weekend
[particle] that was on feather... anyone have any idea what that message means?
audreyt no, audreyt-- for dropping her macbook and breaking it.
16:25 penk left
[particle] (macbook in shower)-- 16:25
agentzh LOL
audreyt that too. 16:26
agentzh audreyt: so on Windows, pugs now requires ActivePerl 5.8.8? 16:27
i think it's a bit unacceptable for most win32 users.
5.8.8 is the latest version, you know. ;-)
audreyt agentzh: no, it requires 5.8.8 for perl5 embedding 16:28
which is enabled by default
but yeah, it's sad. but between that and not emabling perl5 embed 16:29
I think I want perl5 embed.
agentzh audreyt: i see. :)
dada I don't think it's a big problem
agentzh indeed, perl 5 embedding is very useful.
16:29 Eimi joined
dada dumb and lazy win32 users are not going to use pugs anyway 16:29
and installing ActivePerl 5.8.8 isn't that hard 16:30
agentzh dada: but installing half of the cpan *is*
audreyt er, you can upgrade activeperl without reinstalling half of cpan
5.8.x is binary compatible
dada at least if you already have 5.8.something
agentzh audreyt: details?
audreyt (except for 5.8.1 which I prefer to conveniently forget)
dada and if you have 5.6.something, you're really _too_ lazt 16:31
lazy
agentzh by coping over lib/ and site/?
audreyt agentzh: simply install activeperl on top of an existing installation
without uninstalling the old
that's all I think
dada it is
audreyt copying over site/ would work too.
agentzh auderyt: oh, thanks!
audreyt :)
[particle] what's expected passing rate for pugs tests? 16:34
i see *a lot* of failures
audreyt about 500ish failures at this moment I think 16:35
[particle] audreyt: are you running 9m test on osx or win? 16:37
audreyt osx parallel=2
if you "make fast" the test is going to take longer. 16:38
[particle] ok. i wonder because i find win tests take much longer, since creating processes on windows is expensive
also, because i make fast.
i'll have a look in another 20mins then :)
audreyt justatheory: yes, I did. unfortunately I don't have much to contrib 16:41
except that | and & looks more attractive than && and || indeed
(bbiab)
justatheory audreyt: Bummer. You're the mistress of the overloading tricks!
audreyt: I'm fine with & and |, but now I want to overload =~ and !~. 16:42
audreyt that is trickier; though perl5/re-override* in pugs tree shows how. 16:52
but at some point you should give up and do it via Module::Compile or Filter::Simple::Cached :)
agentzh audreyt: while running "nmake", i got "Undefined subroutine &main::UpdateHTML_blib called at -e line 1." 16:54
justatheory audreyt: I don't know those modules; I'll have to take a look. Thanks. 16:55
agentzh audreyt: now i'm at 5.8.8.
audreyt agentzh: nopaste the whole log?
or the parts around it, maybe
agentzh audreyt: okay
pasteling "agentzh" at 210.22.200.67 pasted ""nmake" failure" (93 lines, 3.4K) at sial.org/pbot/19546 16:56
16:58 fglock_ joined
agentzh audreyt: fyi, "nmake fast" works fine and i've built pugs already. unfortunately, "nmake install" and "nmake" don't work. 16:59
audreyt nopaste your Makefile?
agentzh not sure if it's a 5.8.8 on 5.8.7 problem. ;-)
audreyt: okay
audreyt: it's huge. 17:00
:)
pasteling "agentzh" at 210.22.200.67 pasted "my Makefile" (1875 lines, 126K) at sial.org/pbot/19547
agentzh audreyt: sorry, i can't hold on any more. 1:02 AM here...i've got an exam the next morning (well, actually *this* morning). :) 17:05
good night & 17:06
17:06 agentzh left
dada bye people 17:09
[particle] okay, make test finished 17:13
audreyt fixed.
svnbot6 r13124 | audreyt++ | * defeat htmlifypods as it broke build for agentzh++.
[particle] 17 tests and 1157 subtests skipped.
Failed 107/578 test scripts, 81.49% okay. 388/9095 subtests failed, 95.73% okay.
audreyt cool 17:16
how long did it take?
17:18 araujo joined
TreyHarris audreyt: do you agree with Limbic_Region that my 300+m smokes on darwin-ppc (1.5 GHz G4) indicates something is wrong with my installation? 17:18
audreyt: glad you enjoyed my story, btw :-)
[particle] took ~60mins
TreyHarris well, the one last night was "only" 231 minutes.... 17:19
is the smokeserver working? 17:20
ah, spoke too soon
17:20 rindolf joined
rindolf In Moose, how do I set an 'ro' attribute? 17:21
Assuming I can.
[particle] rindolf: C<<writer => '_ro'>> and C<sub _ro{die "cannot set read-only attribute"} # wild guess 17:24
audreyt TreyHarris: GHC 6.4 or 6.6? 17:25
rindolf: if you don't say "is rw" it defaults to ro, no?
17:25 mauke_ joined
TreyHarris 6.4. I'm willing to try 6.6 if anyone can give me a roadmap to get it built on darwin-ppc, but on #haskell they said that's still a work-in-progress 17:26
audreyt nod. don't worry about it for now, I guess
I don't think anything's wrong with installation 17:27
[particle] rindolf: sorry, i was looking at the Moose::Meta::Attribute docs
TreyHarris audreyt: ok, thanks :-)
audreyt GHC is known to be less optimized for ppc than for x86; 6.6 should remedy this somewhat for pugs
but it's nothing to worry about :)
...and now I need to sleep 17:28
good night and see y'all tomorrow *wave*
TreyHarris sweet dreams :-)
[particle] eew, make is really broken on win32
but make fast works fine
audreyt pasteling: even with r13271? 17:29
and reran Makefile.PL?
s/pasteling/[particle]/
[particle] i'm at 13124
audreyt I mean 13124 sorry
[particle] k
i don't see a Makefile.pl update there
but i reran it
audreyt and it still fails? 17:30
TreyHarris was wondering how audreyt expected [particle] to run a revision that probably won't exist until Sunday ;-)
[particle] yse, pasting
pasteling "[particle]" at 144.81.84.199 pasted "fail" (174 lines, 6.7K) at sial.org/pbot/19549
audreyt what's wrong with that? 17:32
[particle] nmake -f Makefile all -nologo
audreyt it's a finite loop with 45 iterations
[particle] huh? why?
audreyt one for each in ext/
it's the subdirs target
[particle] oh, that's entirely unobvious 17:33
i'll let it iterate
audreyt I'd say it is. please see if you can somehow make it more obvious
either by prompting or by saying don't panick
or something like that
[particle] k
audreyt but I can't stay up anymore :)
*wave*
[particle] go away! audreyt++
17:40 mauke_ is now known as mauke 17:42 Limbic_Region joined, elmex joined
Limbic_Region grrr 17:42
TreyHarris Limbic_Region: audreyt just confirmed that my installation isn't broken, that GHC 6.4 on PPC is really really slow :-)
what's wrong? 17:43
Limbic_Region my client just went fubar
TreyHarris ACTION CTCP Limbic_Region TICKLE_CLIENT
did that help? :-)
Limbic_Region [particle] - the problem that agentz was referring to with AS and Module::Build is the one I was telling you about regarding MinGW - it should not affect you 17:44
yes TreyHarris - thanks
for those using build 817 (5.8.8) and are using MinGW - either upgrade or ping me for the patch to make Module::Build work again
Limbic_Region continues to backlog 17:45
[particle] - hopefully our project with util/yaml_harness.pl will help improve make test/smoke on Win32 at least 17:46
5.8.8 may be the latest version but how long has it been out - it isn't like it is exactly new or anything 17:47
TreyHarris - you didn't mention your intermitten bus failures to audreyt though when you asked about the speed issues 17:49
Limbic_Region factored that into his assessment
[particle] - make works fine for me on Win32 as I never run make fast (assuming we are talking about nmake here) 17:50
ok, all finished backlogging
[particle] - if you /msg'd me anything I lost it when this clients went toes up
17:56 Eimi joined 18:04 bradleym joined 18:06 bradleym is now known as bradleym|meet
TreyHarris Limbic_Region: ah, good point. she was heading to bed so I was trying to make it brief, but you're right 18:13
Limbic_Region in any case, have you tried setting the number of concurrent tests? 18:14
TreyHarris Limbic_Region: I don't have an SMP
or even a dual-core
Limbic_Region shakes his head 18:15
if your single CPU is not at 100% utilization, running things in parallel is likely to still be faster than in serial
despite it being done via time slicing
TreyHarris oh, no, it's at 100% pegged
Limbic_Region oh, well then - concurrent tests won't help regardless 18:16
TreyHarris the entire 4 hours
:-)
Limbic_Region my machine stays under 10% for the full smoke (30 minutes or less)
TreyHarris i could speed it up if i shut everything else on my system down. it is my main machine after all. but i'm usually not using it much
ah. even "my $x = 3; say $x" pugs the CPU for a moment 18:17
er, s/pugs/pegs/
pugs pegs the CPU :-)
Limbic_Region so those no op spin cycles while it is waiting to hear back from my program could be used to do work for another program
18:20 mugwump joined 18:23 hexmode joined 18:24 hexmode joined 18:38 xerox joined 18:42 gilimanjaro joined 19:03 hexmode joined 19:23 jferrero joined 19:52 hexmode joined
rindolf audreyt: here? 20:08
Limbic_Region she's supposed to be sleeping 20:10
[particle] lr: i'm back! 20:18
svnbot6 r13125 | fglock++ | Pugs-Compiler-Rule - implemented $<KEY> 20:20
20:23 weinig|bbl is now known as weinig
Limbic_Region [particle] - I just wanted to know if you had done anything on the util/yaml_harness.pl 20:28
get you to /msg me your email again (client barfed)
and ask what your make problem was
since I am building fine here without make fast
svnbot6 r13126 | fglock++ | Pugs-Compiler-Rule - ChangeLog update 20:30
[particle] haven't touched it today (production issue) 20:31
Limbic_Region no worries 20:33
[particle] i think i had a rogue ghc.exe process locking a file
making again to see if that fixed it
Limbic_Region I was so enjoying the YAHT that I didn't notice the time
it is officially 3 minutes into the weekend for me so I am headed homewards 20:34
will send test email in a minute - no need to reply
svnbot6 r13127 | fglock++ | Pugs-Compiler-Rule - ChangeLog update 20:36
20:36 davidfetter joined
rodi any other Perl 6 people live/visit in the Washington, DC, USA area? 20:49
err... #perl6 people :)
20:53 larsen joined 21:04 theorb joined 21:05 vel joined 21:17 Limbic_Region joined 21:22 markstos joined 21:23 fglock_ left 22:02 davidfetter left 22:20 Termy joined, thepler joined
gaal audreyt: thanks for the fix. I'm not sure I understand why this works yet... 22:37
22:37 weinig is now known as weinig|bbl
gaal also, how to write the printers for the non-concrete Val variants? 22:37
the two problems are that they are all monadic, and that they are existential. 22:38
PureVal can maybe cheat and runIdentity...
and the unpacking could maybe case over typeOf but I wonder if there's a more efficient way? 22:39
...maybe there needs to be another pretty printer in Eval? 22:40
anymoose it's night time for me. bye! & 22:41
23:06 Aankhen`` joined 23:11 Aankhen`` is now known as Moose`` 23:25 jferrero joined 23:33 justatheory joined 23:41 penk joined
markstos I was just reading about Prelude.pm ( pugs.kwiki.org/?PreludificationCandidates ) and want to make sure I understand the purpose. I suspect they are primarily: 1. To write some of Perl 6 in Perl 6, and 2. To help support multiple implementations, since anything in Prelude.pm can be used by any implementation. Correct assumptions? 23:41
lambdabot Title: PreludificationCandidates - Pugs - A Perl 6 Implementation, tinyurl.com/gaqv8
clkao yes 23:42
markstos Thanks. I'll update the wiki (.../some wiki/) to clarify this. 23:43
I didn't realize until today that any of core Perl 6 was written in Perl 6, although I'd heard of "Prelude.pm" before, I just assumed it was "different". I know "Prelude" is a term used in the Haskell world, but not so much in Perl culture. 23:44
So I had assumed Prelude==somehowHaskell.
nothingmuch markstos: the meaning is borrowed
it's the part that is before every program =) 23:45
good night folks
Limbic_Region TTFN nothingmuch
23:48 NamelessOne joined
[particle] markstos: parrot also uses something like this. PGE rules are written in perl 6 regex syntax 23:57
markstos Neat.
[particle] so the parrot perl6 parser is written as a perl 6 grammar
TreyHarris one point that I'd appreciate more clarity on: audrey suggested to me that earnest efforts to implement large swaths of S29 should be done in separate, class-based files ("e.g., src/Num.pm" was what she said). i don't understand how Prelude fits together with such other peices
would Prelude use the others? would we have a 'Core.pm' that would use the others, but only Prelude would be used by default? 23:58
markstos TreyHarris: Have you fond this page? It lists candidates for preludification. 23:59
pugs.kwiki.org/?PreludificationCandidates
lambdabot Title: PreludificationCandidates - Pugs - A Perl 6 Implementation, tinyurl.com/gaqv8
TreyHarris i know the overall intention is for base Perl 6 to be much smaller than core Perl 5

Powered by Cro and the Raku® Programming Language.

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

Thank you!