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 2006-08-29 05:26:57 -0700]
Set by Akwa|user on 14 September 2006.
00:01 Psyche^ joined
ingy nothingmuch: ping 00:04
00:13 Psyche^ is now known as Patterner 00:18 frederico joined 00:20 hikozaemon joined 00:29 justatheory joined, nekokak joined
TreyHarris dduncan: so you're not happy with any(@x) ~~ { .condition }? are you generally unhappy with smartmatch semantics lacking explicit identifier versions for particular cases, or just in this one specific case? 00:30
00:30 Psyche^ joined
dduncan unless the syntax of any() is being overloaded here to not make a set/junction, the problem I have is that it looks like it would make a set, which means it won't work for what grep() does ... 00:32
grep() works with ordered lists that can have duplicates, and returns ordered lists; a Junction/Set is not ordered and has no duplicates
but if any() etc is just overloaded syntax that doesn't make a junction, and it is easy to understand as such, then I don't have a problem with the above syntax 00:33
00:33 MacVince joined, Psyche^ joined
dduncan or alternately, if any() etc is not supposed to replace grep, and just does set logic, then I also have no problem with it, as it is doing the set ops tha it looks like its doing 00:34
and I prefer it just does what it looks like its doing
that is all 00:35
TreyHarris well... 00:36
junctions are *not* equivalent to sets, AFAIK. they are individual values that are composites of other values 00:37
so duplicates *should* be preserved
the lack of ordering is true, except than any() in list context has been specifically overridden to preserve order 00:38
00:39 bsb left 00:40 mauke_ joined
dduncan well, Junctions have been defined as being Sets with extra behaviour 00:41
TimToady junctions are a linguistic feature, not a mathematical feature, so it seemed sane to give them an interpretation in list context and not just boolean context.
dduncan though it was never made clear what all the extras were, afaik 00:42
TimToady "any" and "all" are borrowed from English, not from set theory.
dduncan but set theory includes those too ... set calculus
SamB the names should make that clear 00:43
00:43 Psyche^_ joined
dduncan set calculus includes any, all, one, none, etc 00:43
SamB if they were borrowed from maths, they'd have funky symbols!
TimToady so?
justatheory waves to TimToady
dduncan I'm jus saying that they aren't English-only
even if that was your influence 00:44
so I'm glad that they work both ways
TimToady I don't care if set theory has borrowed my wife--that doesn't mean that I'll take set theory's definition of my wife...
SamB ... I would be kinda mad at set theory if it borrowed *my* wife. If I had a wife.
justatheory Or your word that nothing happened with your wife
er, its word
00:45 Psyche^_ is now known as Patterner
TimToady actually, my wife is rather fond of set theory, come to think of it... 00:45
dduncan afaik, the english definition and set definition of those terms is the same, or close enough, so it doesn't matter where they came from
TimToady and I suppose by some definition or other she has been in union with set theory. 00:46
justatheory lol!
dduncan fact is, what we have something that works well
00:46 justatheory is now known as SetTheory
SetTheory Yeah baby! 00:46
00:46 SetTheory is now known as justatheory
TimToady I guess the kid would have to be named Venn... 00:47
justatheory Let me draw you a picture 00:48
dduncan I think there's a unicode symbol for it
TimToady well, speaking of She Who Must Be Obeyed, she just called me and told me to come to dinner. 00:49
so I will go and intersect with her. &
dduncan like all() === all() does 00:50
00:50 mauke_ is now known as mauke
dduncan and I think any() === any() is a union by contrast, or something 00:51
but whatever
or the latter is the intersection 00:52
00:53 bsb joined 01:18 weinig|bbl is now known as weinig 01:24 xdg joined 01:43 hexmode joined 01:45 ludan joined
ludan hi 01:46
svnbot6 r13432 | audreyt++ | * Vivify global syms such as @*FOO and $*FOO. This made 01:50
r13432 | audreyt++ | ./pugs -CPIR work again.
01:54 weinig is now known as weinig|bbl 01:57 buubot joined 02:17 Psyche^ joined 02:23 Psyche^_ joined 02:32 Psyche^_ is now known as Patterner 02:33 cventers joined
cventers hello, i'm doing a Perl 6 article and I would both like to ask a question and see if anyone is interested in reading a prerelase version of said article to make comments about anything I may have missed or misrepresented 02:34
if anyone is interested in looking over the prerelease article feel free to mail me at [email@hidden.address] 02:41
02:47 mako132_ joined
audreyt I'm interested :) 02:48
02:58 markstos joined 03:05 f0rth joined
svnbot6 r13433 | audreyt++ | * unbreak the build 03:08
ingy hi audreyt 03:10
cventers is away: will return in a bit. 03:16
TimToady still seems broken: pugs: user error (*** Undeclared variable: "$fatal::DEFAULT_FATALITY" at <prelude> line 127, column 5-33) 03:34
MacVince !eval my @a = ([1, 2, [3, [4, 5], 6], 7], 8, [9, 10]); say ~@a 03:55
!eval my @a = ([1, 2, [3, [4, 5], 6], 7], 8, [9, 10]); ~@a
svnbot6: eval 2+2
oh well.. 03:56
zgh ?eval say 'foo'
03:56 evalbot_r13431 is now known as evalbot_r13433
evalbot_r13433 OUTPUT[foo ] Bool::True 03:56
zgh gotta start with '?'
04:00 weinig|bbl is now known as weinig 04:15 weinig is now known as weinig|zZz 04:18 kanru joined
svnbot6 r13434 | audreyt++ | * Oops, unbreak for real 04:36
TimToady now here's a cute one: 04:45
make[1]: *** No rule to make target `lib/.CGI.pm.swp', needed by `pm_to_blib'. Stop.
I think something was a little overaggressive with the globbing there...
make[1]: *** No rule to make target `Spec/.Functions.pod.swp', needed by `pm_to_blib'. Stop. 04:49
MacVince .swp? Those are vim temp files. 04:50
TimToady yep. 04:51
TreyHarris i'm still smoking r13336. five hours now... my prior smoke of r13324 too just two hours
s/too/took/ 04:52
TimToady you sure it's not looping?
TreyHarris nope, pretty steady progress
it's in t/xx-uncategorized now, so almost done
it could have been lack of memory... i had a browser crash a bit ago, and in my experience the browser often start chugging memory prior to a crash. 04:54
s/start/starts/
can't type tonight
TimToady you really clear back at 13336, or did you mean 13436? 04:55
TreyHarris hm... no, i'm at r13427... looks like i'm having the problem somebody mentioned earlier about pugs -v reporting an old result 04:57
but of course, that's what smoke reports.... *sigh* 04:58
guess the past five hours--just now finished--have been in vain
TimToady finally got through the make to actually start smoking r13434 myself.
I generally always do a realclean before a smoke, so I never saw the version problem. 04:59
TreyHarris anyone recall what the solution to the wrong version problem was? i'm not finding the right keywords to conjure it up
yeah, besides make realclean... make usually takes an hour for me, make realclean two 05:00
er, make following realclean 05:01
05:02 agentzh joined
TimToady look for "touch" from dduncan 05:03
dduncan yes
try "touch util/version_h.pl"
then "make" again
TreyHarris ah, here it is: touch util/version_h.pl, from dduncan++
TimToady I expect "rm version.h" would work too...
TreyHarris oops. i should scroll back down to the end of scrollback before typing ;-) 05:04
agentzh @tell audreyt i don't know how to reply malon's p6c mail: www.nntp.perl.org/group/perl.perl6.compiler/1511 maybe you're willing to help? :)
lambdabot Consider it noted.
dduncan well, the problem is in whether version_h.pl is invoked at all, not its contents
TreyHarris dduncan: danke
agentzh version_h.pl is significant to smartlinks.pl, i think. 05:05
dduncan I don't know the problem, except to speculate that something in Makefile.PL, which queries svn info, isn't doing the corresponding bit for svk
TreyHarris i use svn
TimToady there isn't a version.h to rm...
dduncan I use svk
agentzh me too
dduncan maybe you mean pugs_version.h ?
TimToady if that's what it recreates.. 05:06
TreyHarris in that case, i guess it's broke regardless of the VC client you're using
mugwump_ version_h.pl special cases svn and svk
TreyHarris well... would it be valid for me to edit my smoke.html to put in the correct version number and upload it anyway?
and smoke.yml...
agentzh that is not elegent... 05:07
agentzh backlogs due to lack of context.
TreyHarris agentzh: no, but i hate to waste a five-hour smoke, particularly as if i'm going to make again i'd like to make the current rev
agentzh TreyHarris: i think that maybe okay. 05:08
smartlinks.pl extracts rev number from tests.yml.
editing .html is fine too. i think. 05:09
dduncan mugwump, the contents of version_h.pl aren't at fault ... its whether that script is called at all that is the problem
TreyHarris agentzh: really? tests.yml is two weeks out of date!
TimToady smoke.yml
agentzh ahh, it has been renamed.
sorry, i didn't know. :)
TreyHarris ah 05:10
agentzh anyway, the yml filename is passed to smartlinks.pl via a commandline option
i didn't hard-code the filename.
TreyHarris i wish i could get a handle on the build process, i'm flummoxed by it. when i make install, it takes 30 minutes or so and rebuilds quite a chunk even though i haven't done anything since the make but a smoke... 05:11
agentzh i agree
"make fast" is more friendly
it seems that the smoking mechanism doesn't call version_h.pl correctly? 05:12
TimToady I gave up on make install, and instead put a symlink from /usr/bin/pugs into my build dir. 05:13
dduncan "make" in general doesn't call version_h.pl correctly
the smoke mechanism gets the number directly from Pugs
so the smoke will say whatever pugs -v says
agentzh from .h?
dduncan this isolates the matter of querying VC to just "make" time
I know because I made it do that myself
agentzh i'll try to fix that.
TreyHarris TimToady: aye, but when there's regression, i like having an installed pugs that i can continue using for whatever has regressed...
TimToady there is that... 05:14
05:14 BooK joined
TreyHarris well, i'm getting a faster machine this week. not tremendously faster, a core duo 1.8GHz... but still loads better than this 1.5GHz G4 ppc.... 05:15
dduncan what machine is that?
TreyHarris and with 2GB of real RAM (this one is 1.25GB), so that should help too. 05:16
dduncan: a mac mini
dduncan I see
I'm on a dual 1-Ghz G4 myself, which probably about the speed of your current machine
TreyHarris i bet your smokes are faster
dduncan 768M of RAM though ... could use more, particularly when building some parts of Pugs 05:17
05:17 coumbes joined
TreyHarris especially if you have concurrent set to 2 05:17
dduncan I recall about 2.5 hours
agentzh it seems that $version_h doesn't get updated if one is using svk.
TreyHarris agentzh: i'm using svn
agentzh i'll try a fix.
TreyHarris: you're having the same problem? 05:18
TreyHarris yeah, my old iMac G4 800Mhz (bought the day they came out in Feb. 2002) finally gave up the ghost this past weekend. i thought about a new imac, but i'd really like to have an external monitor to kvm with...
agentzh: yes
dduncan hm, was the smokeserv reset ... I don't see the older smokes anymore
agentzh thinking... 05:19
dduncan Trey, the smoke used to take around 5h to do on my machine, when Pugs' multi-threading was broken on my machine ... later it was fixed and them my smokes were about 2.5 hours
TreyHarris agentzh: my pugs is reporting r13336, but i just build r13427
dduncan broken meaning that smoke_concurrent > 2 didn't have any effect, with just 1 smoke process at once being run 05:20
TreyHarris s/build/built
dduncan er >= 2
05:20 traecer joined
agentzh TreyHarris: what do you mean by "reporting" 05:20
and what do you mean "just build r13427"
?
TreyHarris TimToady: is it/will it be legal to use unicode arrows for ->, ==>, <==, <=>, etc? 05:21
dduncan reporting = pugs -v
TreyHarris agentzh: pugs -v reports r13336, but i just built r13427
agentzh okay
dduncan Trey, did you try the touch yet?
agentzh fixing
Trey: does touch work for you? 05:22
TreyHarris dduncan: about to. first i'm going to edit smoke.html and smoke.yml. i think perhaps i should s/13336/13427/, but not in the pugs banner, to document this problem? 05:23
dduncan the problem is documented by us talking about it here
you can update your smoke files as if the version was working, since the public looking at them should have more accurate info 05:24
TreyHarris ok
dduncan or if you want to document it, add extra plain text text somewhere in the smoke html, which hopefully won't cause a parse problem on the smoke server
in addition to changing the nums 05:25
TreyHarris already uploaded with a global s///
dduncan good going gone
TreyHarris now i'll try the touch, but i guess i won't know for a bit
TimToady TreyHarris: not unless you declare them yourself, probably.
dduncan you'll know quickly
following the "touch ...", just do a "make", which should just take 1 minute or so 05:26
assuming you haven't made clean
then pugs -v is all the verify you need that it worked
that and, ...
a diagnostic line during "make" should say "compiling version.hs" rather than "skipping version.hs"
TreyHarris dduncan: ah, yes indeed 05:27
dduncan also, around the 6th output line during "make", it should say "version is NNN" or something like that
TreyHarris dduncan: but as i said before, make (for me at least) always rebuilds some things, even if nothing is changed in the interim
dduncan: yes, it did.
so the minimum time reported by 'make; time make' is 20 minutes or so 05:28
dduncan well, "make" is verbose, but it should only take a minute with no changes if you system is like mine
my pugs setup is basically just all the default settings, but some minor changes to config.yml
TreyHarris dduncan: interesting. do you have an up to date ./pugs in your working copy now? 05:29
dduncan I also did a clean checkout just last week
more or less ... checking ...
my pugs -v says r13426
TreyHarris run make again
see if the timestamp on pugs changes. in my case, it would 05:30
i don't have to touch anything but it gratuitously rebuilds pugs
dduncan actually, my pugs is r13433, but reports the above since I didn't touch the file since last time
34 has an Eval.hs change 05:31
since I'm making with that, it'll take a bit longer, starting now...
TreyHarris ah, right. but what i'm interested in is, once you're done, without touching a thing, running make again... 05:32
dduncan will tell ...
TreyHarris oh, that was comparatively fast. only 8m
urgh, pugs -v still reports 13336
dduncan biab 05:33
TimToady pugs or ./pugs?
TreyHarris ./pugs
even though src/Pugs/pugs_version.h contains: "#define PUGS_SVN_REVISION 13427" 05:35
odd.
agentzh TreyHarris: you can fix util/yaml_harness.pl. :) 05:36
s/fix/patch/
TreyHarris agentzh: yaml_harness.pl isn't what creates the -v string, is it?
agentzh TreyHarris: i'm afraid that is intentional. 05:37
or you'll have to rebuild more after your svn update
TreyHarris agentzh: elaborate?
it's fine, i don't mind realcleaning this time. but if i have to realclean *every* time... that is suboptimal. 05:38
dduncan back ... and make finished
so <= 5 min that time
agentzh TreyHarris: i think the real problem is the rev number contained in your smoke reports, right?
i can fix that.
TreyHarris agentzh: no, a perl -i -ape fixed that right up :-) 05:39
er, s/-a/-/
05:39 szabgab joined
dduncan version didn't change just then, a plain make after didn't change it either, but a touch and a make made it updaet 05:41
TreyHarris dduncan: touch of util/version_h.pl?
dduncan that workaround continues to work
and without that workaround, it is incorrect 05:42
agentzh dduncan: made it update every time is easy.
TreyHarris hm. i wonder why not for me?
agentzh *making
dduncan it seems to me that the Makefile is treating version like typical other files where it doesn't remake it if it didn't change ... but version is special and shouldn't be treated that way, and in the past it wasn't 05:43
agentzh gotta run &
05:43 agentzh left
TreyHarris ahh, i just found a possible culprit 05:44
dduncan did you now?
TreyHarris libHSPugs-6.2.12.a was well out of date, almost a week
05:44 traecer joined
TreyHarris i removed it and am re-making 05:44
dduncan brb 05:45
TreyHarris not almost a week... to Sep 15, which was when r13336 was
yes, that fixed it
but that just pushes the problem... why didn't make realize it needed to create a new libHSPugs? 05:46
and of course, this means that my smoke really was invalid...
dduncan Trey, from what I could see, pugs_version.hs contained the old value 05:52
that file wasn't recompiled because it didn't change 05:53
that file is set by version_h.pl
do you know for a fact that Pugs wastn't rebuilt aside from the version number?
from my determination, your smoke is still valid 05:54
TreyHarris the timestamp
and i hate to think that it takes an hour to *not* build a new pugs from source :-)
dduncan if that is so, then we may have multiple problems
or multiple manifestations of the same source problem 05:55
TreyHarris *nod*
dduncan anyway, back to $job for me ...
TreyHarris dduncan: ...
dduncan yes
TreyHarris well, what if I were to svn up now... 05:56
first do a recursive ls -l...
then svn up... then make
and, assuming -v doesn't update, see what had not been updated that one would think should have been?
dduncan you want to see what changed?
TreyHarris might give us clues
dduncan good luck with that
actually ...
if you copy the whole checkout dir first 05:57
TreyHarris seems like the problem is 1) targets are rebuilding when their sources have not, and 2) targets are not rebuilding when their sources have.
dduncan then if you have a utility to compare directories including file contents, ...
you could then see the old versions and new versions where they exist
TreyHarris *nod*
easily done. rsync --dryrun :-)
dduncan I have yet to use any sort of rsync on purpose 05:58
backups are still manual
TreyHarris *grin* 05:59
ok, go $job
i'll poke around.
06:23 marmic joined 06:38 f0rth joined 06:40 mdiep joined 06:43 traecer joined
nothingmuch ingy: pong 06:47
lambdabot nothingmuch: You have 1 new message. '/msg lambdabot @messages' to read it.
nothingmuch @seen putter 06:52
lambdabot I saw putter leaving #perl6 10h 1m ago, and .
nothingmuch oi
07:05 smee joined
smee i kno more about computas than u all im da best hacker eva 07:05
07:09 smee left
webmind then way didn't he stay to finish perl6? ;) 07:11
07:13 prefiks joined
obra because he already did it ;) 07:14
07:27 iblechbot joined
TreyHarris last time he came on, audreyt offered him a commit bit :-) 07:27
nothingmuch @tell putter commits welcome ;-) darcs send to nothingmuch.woobling.org/MO 07:33
lambdabot Consider it noted.
nothingmuch @hug
lambdabot Maybe you meant: msg run
nothingmuch @botsnack
lambdabot :)
nothingmuch ARR! how ye be fairing?
07:33 prefiks left
svnbot6 r13435 | malon++ | smokeserv/smokeserv-server.pl - fix broken links to tests from TAP matrices 07:38
r13435 | malon++ | - Reported by markstos++
r13435 | malon++ | - Note that ext/ links will stay broken, because those files aren't checked out
07:43 traecer joined
Juerd It's Prinsjesdag here today. The day on which the government announces the new budgets, and the queen reads this all. 07:45
07:45 drrho joined
nothingmuch Juerd: here that could constitute as a talk like a pirate day activity 07:48
they are taking more budgets from welfare
and directing them to the army 07:49
so the pilots can have better lounges
Juerd Well, I just can't wait for the Queen to talk like a pirate.
...for more than an hour
nothingmuch =)
Juerd "Yarr mateys of the Staten Generaal!"
("Geachte leden van de Staten-Generaal", "Dear members of the parliament")) 07:51
s/))/)/
08:04 dduncan left 08:11 cookys_ joined 08:12 dduncan joined
dduncan I have a question on ... I think it's called "parameterized types" 08:13
08:14 idiotmax joined
dduncan or "parametric types" 08:15
TreyHarris heh, shoot... i was just writing something about them :-)
08:16 ron__ joined
dduncan right then ... I was thinking that perhaps I have been thinking too complicated about how to make a Tuple type 08:16
then I thought, I don't know if I can do this myself or if it has to be built-in
if I could just say:
my Tuple of { Str $name, Int $age } $foo; 08:17
or there-abouts
the part in the curlies is meant to look sort of like a Signature
on the other hand, its meant to look kind of like a C struct, but without the cruft 08:18
nothingmuch dduncan: classes can receive tytpe parameters
TreyHarris I believe it would be my Tuple[Str, Int]
dduncan they need to be named
nothingmuch class Tuple[::t1, ::t2] { has t1 $fst; has t2 $snd }
it kinda sucks that they're positionals
but i believe it's on the agenda to fix them 08:19
08:19 kanru joined
TreyHarris my Tuple[Str, :name, Int, :age] today I think then 08:19
dduncan part of the thing is, the base Tuple type itself doesn't declare any attributes ... its users do that
TreyHarris though none of this works yet, so 'today' only in the sense of the current state of the synopses
08:20 ruoso joined
dduncan same as users can say "Array of Int" vs "Array of Str" 08:20
nothingmuch so what type is it?
why not just use an anonymous class?
dduncan okay, how do I do that? 08:21
nothingmuch my class { has Str $.name; has Str $.age } $foo;
dduncan I see
nothingmuch i suspect
my $class = class { }; # this you can definitely do
and then you can bind it to a lexical type, i guess
my ::foo := $class;
my foo $var;
that *should* work
but it's only meaningful at runtime, of course
TreyHarris anonymous classes are supported, though their syntax is unspecified. i'd be surprised if my class { .. } $foo did not work though. 08:22
dduncan even if that works, I would need syntactic sugar so that it looks like they are declaring a restricted hash 08:23
what I want is sort of like a restricted hash but that in addition to the names, you declare types for each possible member
which then looks more like a class def 08:24
TreyHarris but, if you're going to have more than one Tuple of the same Tuple type, and you need to support operations on two Tuples iff they are of the same Tuple type, then you'd need to use parameterization, and agreed, the current syntax is somewhat lacking.
dduncan maybe the anons will do what I want though
TreyHarris creation of new containers is currently unspecced as well, and it sounds like you're looking for a new container type. 08:25
dduncan there is a set of operators specific to tuples that these anons would need to have, so naming them as a parameterized class could do that, or I could say anon {...} does Tuple role
I'm hoping to reuse the meta-model functions to do some of the hard stuff
it is possible that a new container type is appropriate 08:26
my main goal here, however it is accomplished, is that the multi-paradigmitic Perl 6 would support the relational data model as one of its paradigms ... which means tuple and relation data types and operators
nothingmuch you could declare a meta object for tuples 08:27
dduncan they don't necessarily have to have the same names as existing implementations
nothingmuch and declare a DSL that inherits the grammar for classes
and simply constructs a different meta object as a result
dduncan well, I would like to know how I do that then
nothingmuch but that's not going to happen soon
dduncan part of my problem is that it appears that the meta-model is unspecced
nothingmuch i have a sketch that was working well for me 08:28
based on moose
which incorperates more than a year of research by stevan and audrey
audreyt wants to add it into pugs
dduncan and being a perfectionist of sort, I'm having trouble trying to make a half-baked version using what features do exist
nothingmuch if that sounds good enough for you to poke with: nothingmuch.woobling.org/MO/
lambdabot Title: Index of /MO
nothingmuch *nod* 08:29
dduncan I want to be able to use the meta-model api, yesterday
nothingmuch me too
that's why i wrote one ;-)
i hope larry likes it
dduncan the ironic thing is, isn't the Perl 6 meta-model already implemented?
isn't Perl 6 using it right now?
nothingmuch several are
not really
dduncan so is it just a matter of exposing the API to perl programs? 08:30
nothingmuch well
pugs' meta model is more of a temporary placeholder
it's optimized for fun
dduncan which one is yours?
nothingmuch that is, getting the max benefit in the least amount of heeadache
mine is just in that URL
it's in perl 5
it's designed to easily work in haskell too 08:31
dduncan one issue is, in my mind, what I require should be technically very simple
TreyHarris well, without creating a new container, and if you wanted it at compiletime, I think you'd have to do something like class Tuple[Array of (Str Pair{::Type})] or some such....
dduncan you used parens right there; what does that signify? 08:32
TreyHarris er, no, it would be a value parameter, not a type parameter, so class Tuple[Array of (Str Pair{::Type}) $tupledef] or some such....
nothingmuch dduncan: this is all highly conjectural
TreyHarris dduncan: just grouping parens, and == nothingmuch
probably unnecessary as i read S09 08:33
nothingmuch this is one of the areas that will probably get a big makeover
dduncan regarding the meta-model API, is that something that @Larry is designing, or is someone else like yourselves doing it officially?
nothingmuch once it's actually implemented
TreyHarris but your tuple would need an array of pairs with type keys and string values
dduncan pretty much
nothingmuch dduncan: no word from @Larry yet 08:34
dduncan er, it is a mapping of attribute names and type names
nothingmuch but once i feel that MO is kinda done, i will propose it officially
dduncan s/type names/type names or other anonymous type defs/
nothingmuch and hopefully @Larry will run it through and extract the parts they agree with and we will at least have something
btw, MO doesn't deal with typing
that's for a separate layer 08:35
TreyHarris unordered? then you could use class Tuple[Str %attribs{::Type}] or somesuch
dduncan is MO a design or does it work now?
nothingmuch that must interpret the meta model
MO works
but it's slow
and desugared
dduncan yes, tuple attributes are unordered by definition
nothingmuch my next step is compiling it down to perl 5
for the named method dispatch type
which is the only paradigm supported by perl 5 ;-)
TreyHarris dduncan: interesting. i was taught that tuples are ordered
dduncan a Tuple is like a key-and-value-type-constrained Mapping or Hash 08:36
TreyHarris how does a tuple differ from a struct then?
nothingmuch TreyHarris: he means something more like a record
dduncan yes
actually, the term tuple is overloaded
in the relational data model, it is a set of zero or more triples: name, type, value
the names are all unique 08:37
and keying is by name rather than by position
a tuple type is a collection of name, type pairs
and values of that type fill in the value parts
08:37 ron__ joined
dduncan sort of like class definitions ... 08:37
nothingmuch goes to excercise
dduncan a tuple type is a class and a tuple value is an object
this said ...
if using the word Tuple is going to confuse people, where some think they are ordered, we could use another word ... maybe Record instead 08:38
it doesn't matter so much to me what it is named, as long as something with the right API and a descriptive name exists
TreyHarris well, wikipedia, the mother of accuracy, says tuples are ordered in the first sentence ;-) 08:39
dduncan one minute ...
that's domain specific ... some domains call tuples ordered, some don't
so perhaps Tuple isn't the best name, considering both of those meanings exist
TreyHarris but even if it is an overloaded term, I certainly learned the term as meaning ordered and until today wasn't aware that there was contention on the matter. but your argument in re set theory versus English could apply here :-) 08:40
yes :)
dduncan mind you, even the ordered version comes from math
what about "Record"? ... does that suggest that fields are ordered? 08:41
TreyHarris but yes, the type parameter for whatever you call it would be that you'd need a mapping of types to strings (or some other symbolic thing)
dduncan elements of tuples are called attributes, at least in the relational model 08:42
TreyHarris dduncan: i think record is such an incredibly overloaded word that you can usurp it for any typed attribute relationship you might like and no one could reasonably argue :-)
dduncan that said, I do still prefer the most specific name possible that is applicable
I'm trying to search wikipedia now, but it seems to be stuck 08:43
TreyHarris well, then, it might be clearer to define a type, ThingieAttribute, containing a Type and a Symbol (or Str or whatever), and then say that Thingie (I'll leave it to you to decide a name) takes a Bag of ThingieAttributes
heirarchical typing is a PITA in every language that supports it. 08:44
AFAIK 08:45
zgh that's a quotable quote 08:47
TreyHarris zgh: which? :-)
dduncan and Perl 6 supports it already
eg, Foo of Bar of Baz
zgh 01:45 < TreyHarris> heirarchical typing is a PITA in every language that supports it.
TreyHarris well, that's not so bad, but (Array of Bar) Foo of (Array of Baz) Bat and such.... 08:48
zgh and perl6's support is more fun than most
dduncan yes
TreyHarris zgh: i'm liking it so far, but i worry about it. trying to finish a writeup of one of my concerns in fact 08:49
zgh TreyHarris: look forward to reading it 08:50
dduncan Trey, post that thing you're writing to p6l or something when you're done 08:51
and btw, even if it isn't sanctioned yet, I would be happy to use even an unofficial metamodel API soon, so there's something to play with
anyway, I'm off & 08:52
zgh Moose++
lumi Haskell does the higher-order typing nicely
dduncan I want to play with it in perl 6 though
TreyHarris lumi: I need to relearn Haskell. I learned it immediately after Gofer and the two are quite muddled in my mind 08:53
where is YAHT, btw? i went looking for it the other day and didn't find it, except for the copy in the pugs repo which appears to be truncated
lumi There isn't a link in READTHEM? 08:54
08:54 ron__ joined
TreyHarris lumi: there is, but it is 404 08:54
perlbot: YAHT 08:55
perlbot Yet Another Haskell Tutorial -- www.isi.edu/~hdaume/htut/
TreyHarris yes, that's broken
lumi www.cs.utah.edu/~hal/publications.html
(He moved, apparently)
pub.hal3.name/#daume02yaht 08:56
TreyHarris perlbot: relearn YAHT www.cs.utah.edu/~hal/publications.html
08:56 Sal joined
TreyHarris perlbot: YAHT 08:56
perlbot Yet Another Haskell Tutorial -- www.isi.edu/~hdaume/htut/
TreyHarris hrm.
perlbot: relearn YAHT as Yet Another Haskell Tutorial -- www.cs.utah.edu/~hal/publications.html 08:57
perlbot relearned entry for YAHT
TreyHarris perlbot: relearn YAHT as Yet Another Haskell Tutorial -- www.cs.utah.edu/~hal/docs/#daume02yaht 08:58
perlbot relearned entry for YAHT
08:58 foo\ joined
lumi perlbot: YAHT 09:00
perlbot Yet Another Haskell Tutorial -- www.cs.utah.edu/~hal/docs/#daume02yaht
svnbot6 r13436 | lumi++ | YAHT moved, fixed URL
TreyHarris lumi: hah, I just conflicted trying to make the same change but using the other URL :-)
09:01 dduncan left
TreyHarris dduncan: looks like in Haskell, a tuple is ordered 09:02
09:02 cookys joined
lumi TreyHarris: I think he means that URL to be permanent 09:02
TreyHarris perlbot: relearn YAHT as Yet Another Haskell Tutorial -- pub.hal3.name/#daume02yaht 09:03
perlbot relearned entry for YAHT
zgh boggles at 192 pages
that's a lot of tutorial
mugwump_ that's ok, it means the slope's not so steep ;) 09:04
zgh i really liked haskell for the month or two that i knew it, but that was years ago and now it's incomprehensible again ^_^
zgh reads & 09:05
TreyHarris well, my first formal programming course was in Haskell, so it should be like riding a bicycle, but strangely it's not.
09:06 ron__ joined, drrho joined
zgh wow the descriptions of haskell's goals sound amazingly like perl6 09:07
's goals
lumi Perl 6 is all about ideas that enjoy wide consensus :) 09:08
TreyHarris another random typing question: does every type encompasses undef? or is it necessary to write (--> Int|undef), etc.? 09:12
ruoso wonders if there is some class/module modifier to tell it should be read-only (in the terms of a shared library) 09:14
TreyHarris you mean closed? 09:15
ruoso hmmm 09:16
TreyHarris S12:1521: "There is I<no> syntax for declaring individual classes closed or final. The application may only request that the optimizer close and finalize unmarked classes." 09:17
ruoso It makes more sense to think as Roles actually 09:18
actually, roles works just like shared libraries
TreyHarris but one can instantiate a role as an anonymous class. so the above would apply to roles as well, i think. 09:19
ruoso TreyHarris, but I think you don't have a way to change the Role itself 09:21
you can apply other roles and traits to an object
or even create an anonymous class that "does" a role
TreyHarris role Foo { ... }; role Foo is also { ... }
ruoso can you use "is also" in a role? 09:22
TreyHarris no, sorry, you're right 09:23
ruoso hmmm..
TreyHarris roles are closed at compile time
ruoso having weird ideas
If this constraints apply to roles, 09:24
09:24 idiotmax joined
ruoso one could make standard shared libraries from them 09:24
and use them wherever he wants
does it makes sense? 09:25
of course this would mean linking to the vm
TreyHarris i need more context to know what you mean.
ruoso It's possible to create Perl 6 shared libraries 09:26
that loads as any other shared library
and are actually shared among processes 09:27
TreyHarris why would that preclude runtime extension?
a perl 6 shared library wouldn't be ELF or dylib or something like that, it would be something special to Perl 6. 09:28
09:28 ivas joined
ruoso trying to think better and ellaborate more on that 09:30
09:31 zgh joined
ruoso a shared library is just a ELF that declares some symbols with pointers to the code to be run as these symbols 09:33
09:35 zgh joined 09:37 idiotmax joined
ruoso still thinking... 09:38
09:52 penk joined 10:13 zgh joined 10:30 ptc joined 10:41 b00t joined 10:49 jferrero joined 10:57 agentzh joined
agentzh i'm fixing the pugs -V problem. 11:03
11:03 TimToady joined
agentzh it seems that ghc won't compile Pugs.Version even if pugs_version.h has been updated. 11:05
so i'd make version_h.pl explicitly touch src/Pugs/Version.hs. 11:06
this should work for TreyHarris.
for svk users, i'd set $svn_entries to 'force_run' which is a phony target. 11:07
so pugs_version.h will be updated at every "make"
(for svk only)
for svn users, .svn will still be honored. 11:08
11:08 chris2 joined
agentzh unfortunately i don't use svk, so i can't effectively test the svk part. 11:10
11:11 iblechbot joined
lumi agentzh: I use svk, how would I test this? 11:12
agentzh lumi: i'm still investigating the svn part. (there seems to be some problem.) i'll ping you if i'm ready. :) 11:13
lumi Okay 11:14
agentzh .o O { testing the pugs build system is really tricky } 11:15
11:18 elmex joined 11:22 DucK_ joined 11:23 cognominal joined
agentzh done! 11:29
lumi: can you help me now?
lumi agentzh: Yep, tell me who to shoot 11:30
agentzh after i commit the change, please don't rebuild your pugs.
after "svk pull", simply run "Makefile.PL".
11:30 christopher joined
agentzh committed. 11:31
svnbot6 r13437 | agentz++ | * attempt to fix the rev number mismatch in pugs -v
agentzh after "perl Makefile.PL"
run "make fast"
make sure you see the line "Current version is 13437" 11:32
and the line "===> touching D:/projects/Pugs2/util/..//src/Pugs/Version.hs"
(or something like that)
note that you can't test twice. 11:33
only the first time is meaningful.
lumi Yes and yes 11:34
agentzh please also tell me if you see the line "Compiling Pugs.Version"
lumi pugs_revision: 13437
Also yes
agentzh and the rev number in pugs -v?
lumi 13437 11:35
agentzh great.
lumi agentzh++ :)
agentzh can you run "make fast" again.
?
lumi Yes, what should I see?
agentzh this time make sure you see the line "Skipped Pugs.Version" instead of "Compiling Pugs.Version" 11:36
oh, i mean "Skipping"
lumi Not writing src/Pugs/pugs_version.h because 13437 == 13437
agentzh aye
lumi And it didn't show any of the files it's skipping for some reason, and thus didn't shoe "Compiling Pugs.Version" either 11:37
agentzh and also no "touching"?
lumi Oh, because I have 6.5
Yes, no touching
agentzh oh, i see.
lumi s/shoe/show/
agentzh so it works now!
yay!
lumi Hooray! :)
This should result in correct smokes, right? 11:38
agentzh i'm looking forward to more test reports.
lumi: sure
the smoker reads rev number from pugs -v
if anyone still need the "touch util/version_h.pl" or "make realclean" workaround to get the correct pugs version, please let me know. 11:40
*needs
lumi: thanks for the help. :) 11:42
lumi++
agentzh looks into the Judy win32 rebuilding problem now. 11:44
11:56 Limbic_Region joined 12:01 discordja joined 12:10 xinming joined 12:19 ofer0 joined 12:24 iblechbot joined
ptc #join parrot 12:27
Limbic_Region ptc try s|#|/| 12:28
although you may be disappointed
#parrot is not on freenode
you want irc.perl.org
salutations all btw
kolibrie Limbic_Region: salutations 12:29
12:32 lumi joined 12:33 agentzh joined
agentzh writing Makefile for nmake is a nightmare... 12:34
12:34 zgh joined
agentzh i'm making full use of my perl 5 skills to do it right. :( 12:34
Limbic_Region agentzh - by hand? Or are you writing code to write a makefile
agentzh 60% is generated while 40% is written by hand. 12:35
Limbic_Region oh - if you are writing code to generate the Makefile then there should be some nifty tricks you can pick up by looking at how Pugs does it
agentzh 100% automated seems impossible.
Limbic_Region: sure 12:36
Limbic_Region in fact, the generated Pugs Makefile doesn't even require nmake.exe 7.x (I accidently had a $ENV{PATH} problem and used 1.5)
agentzh Limbic_Region: yes, i'm also writing nmake 1.5-compatible code. 12:37
Limbic_Region short lines are key there
agentzh is just trying to understand the Judy compilation flow.
12:37 discordja joined
agentzh Limbic_Region: really? i didn't know. :) 12:38
so i need to use "\" to split long lines?
i didn't have nmake 1.5 so i can't test it. :)
Limbic_Region agentzh - I am not even sure that will work though audreyt would know for sure
agentzh s/didn't/don't/ 12:39
okay
Limbic_Region she is the one that fixed the Pugs Makefile for me due to the long line breakage IIRC
agentzh ah 12:40
Limbic_Region does this have anything to do with Makefile::Parser ?
agentzh to my surprise, pugs' Makefile on my machine contains a lot of very long lines. :)
Limbic_Region: nah 12:41
i'm just wanting to offer a Makefile.win for judy.
so we win32 guys won't need to build judy at very make if we have a Makefile. 12:42
Limbic_Region agentzh - so what are the major differences then between a nmake compatible Makefile and say - dmake or ming32-make or msys make?
agentzh i planned to rewrite Makefile::Parser completely this summer. unfortunately i sidetracked to pugs. :) 12:43
Limbic_Region s/un//
agentzh nmake is much more stupid.
there's very few magic in nmake that i can use.
i'm not familiar with dmake. not sure if it's better. 12:44
gmake is very very powerful compared to nmake.
Limbic_Region well, IMO, the solution would be to remove the dependency on nmake and use another freely available make on Win32 12:46
but that's just my opinion
agentzh well, my Makefile.win32 should also work with other versions of make. 12:47
its syntax very very simple.
Limbic_Region well, good luck 12:48
agentzh *is
thanks
12:52 ludan joined
ludan ciao 12:53
13:12 baest joined 13:19 dakkar joined 13:24 fglock joined
agentzh yay! now we have Makefile.win32 for Judy! 13:24
This Makefile saves 1/3 of the running time of "nmake fast" if there's no change in the pugs tree. 13:25
13:25 crem joined 13:28 weinig|zZz is now known as weinig|away
svnbot6 r13438 | agentz++ | [judy/Judy-1.0.3/src/Makefile.win32] 13:28
r13438 | agentz++ | - added this Judy Makefile for Win32 users.
r13439 | agentz++ | [util/build_pugs.pl]
r13439 | agentz++ | - the pugs build system now uses Judy's Makefile.win32
r13439 | agentz++ | instead of .bat.
13:36 discordja joined 13:46 BooK joined 13:50 chris2 joined
svnbot6 r13440 | agentz++ | [third-party/judy/Judy-1.0.3/src/hs_build.bat] 13:50
r13440 | agentz++ | - added comment that this file is no longer used by
r13440 | agentz++ | the pugs build system.
fglock mm - I believe t/xx-uncategorized/rule_recursive.t is wrong - the first choice in / <null> | x <r> / is <null> 13:51
13:53 agentzh joined
svnbot6 r13441 | agentz++ | [third-party/judy/Judy-1.0.3/src/Makefile.win32] 13:59
r13441 | agentz++ | - added missing items to its `clean' target.
r13442 | agentz++ | [util/build_pugs.pl]
r13442 | agentz++ | - added /nologo to nmake calling.
audreyt fglock: I noticed too, it should be corrected
fglock ok 14:00
svnbot6 r13443 | agentz++ | [Makefile.PL] 14:02
r13443 | agentz++ | - added the `judy_clean' target to the Makefile to be 14:03
r13443 | agentz++ | generated. judy_clean invokes "make clean" on *nix and
r13443 | agentz++ | "nmake /F Makefile.win32 clean" on Win32.
r13443 | agentz++ | - made `judy_clean' a dependent of the `realclean'
r13443 | agentz++ | target.
r13443 | agentz++ | (I'm sure cmarcelo will be happy with this commit. :))
r13444 | agentz++ | [Makefile.PL] 14:06
r13444 | agentz++ | - renamed the `judy_clean' target to `cleanjudy'. :)
agentzh anyone on *nix willing to test Makefile.PL for me?
simply run "perl Makefile.PL && make cleanjudy" and paste the outputs here. 14:07
(i mean via nopaste)
i can only test the win32 part of the new patch. 14:08
14:10 ofer0 joined
agentzh thinks #perl6 is kinda inactive at this moment. :) 14:11
i'm pleased to see the pugs build system becomes better after tonight's work. whee! 14:13
14:14 thepler joined
agentzh i'm not sure if it's really worth my time because audreyt has said the build system will experience great refactoring once GHC 6.6 is out. 14:15
but anyway...
agentzh now tries to find something else to do. 14:17
okay, 22:17 PM here. end of day for me & 14:19
14:19 agentzh left
ajs TreyHarris: I disagree with both the assertion that Perl 6 code as shared lib would prevent runtime modification AND that Perl 6 as shared lib would not be an elf or similar shared object. There are lots of ways to do both. 14:19
lambdabot ajs: You have 1 new message. '/msg lambdabot @messages' to read it.
ajs Thank you lambdabot! ;) 14:20
14:20 agentzh joined
agentzh @tell TreyHarris the pugs -v rev number problem ought to have been solved. 14:21
lambdabot Consider it noted.
agentzh @tell dduncan the pugs -v rev number problem ought to have been solved.
lambdabot Consider it noted.
ajs agentzh: I'll run it for you
agentzh @tell TimToady the pugs -v rev number problem ought to have been solved.
lambdabot Consider it noted.
ajs tell me how to nopaste 14:22
agentzh ajs: nice!
kolibrie ajs: see topic
agentzh sial.org/pbot/
lambdabot Title: sial.org Pastebot - pasteling
ajs I have to re-update. I was hunting through revisions for a bug, so I need to get a clean update first... just a few
agentzh ajs++ 14:23
ajs Hmmm... gnupg is running a logo contest. Makes me think about a pugs logo contest. That would be an interesting item for just about the time we're ready for wider beta. 14:24
agentzh hehe 14:25
that's a funny idea
ajs I can just see someone doing a pug dog with a hump on it's back and dollar-signs for eyes.... 14:26
agentzh lol
14:26 penk joined
ajs damn, I'm getting this "object of the same name already exists" problem where I have to wait for svn to die for 30 seconds, delete something and re-run 14:27
14:28 buetow joined
agentzh ah 14:28
14:28 zgh joined
ajs It's an iterative thing... will take me a few more min 14:29
14:29 hexmode joined
agentzh no problem. :) 14:29
i can wait.
ajs Pastebot is nice. 14:30
What are the current efforts with respect to handling Perl 5 code? Are there multiple? 14:31
agentzh you mean p5 embedding or v6.pm? 14:32
ajs I mean Perl 5 code as parsed by Perl 6 compilers. 14:33
agentzh doesn't quite understand.
ajs use v6; my $x = 1; use v5; $x=~s/(.)\1//g; use v6; say $x;
or whatever. dumb example, but you get the point 14:34
agentzh oh, i'm not aware of such kind of things in pugs.
iirc, v6.pm can handle use v5.
but not sure.
ajs Ok, I thought someone was writing a p5 parser/runtime since PONIE was kaput
agentzh fglock: confirm it?
fglock agentzh: yes 14:35
agentzh: but v6.pm doesn't compile p5 code - it passes it to perl5 14:36
ajs I really want to know what makes svn lose it's brain and decide not to update a subdir unless I manually blow it away
agentzh got it
ajs fglock: Yeah, it seems to me that we need a pure Perl 6 parser/runtime for Perl 5 at some point, though that's a huge job, and figuring out how to move over XS would be interesting. 14:37
agentzh fglock: btw, i'm so glad to see you are implementing missing features in PCR.
.o O ( only perl5 can parse perl5 ) 14:38
fglock agentzh: thanks!
ajs agentzh: Hence the proposal that Perl 6 parse Perl 5 ;)
agentzh fglock: thank _you_!
fglock ajs: I think the p6p6 compiler can be reused to compile perl5 - but it would need a lot of tweaking 14:39
agentzh ajs: that can be very hard i'm afraid. :)
ajs I was looking at it earlier. I would never have wanted to write lex/yacc from scratch for P5. It's WAY too context sensitive. However, P6 rules make a lot of that easier to cope with.
fglock the perl6 parser is actually much simpler than the perl5 parser
ajs fglock: yes! 14:40
14:40 justatheory joined
ajs The problem with P5 is that you have to read the P5 source code for the spec, and it's often heuristicly defined. 14:42
avar p5 can't be parsed with a plain lex/yacc combo
ajs cd third-party\judy\Judy-1.0.3\src && nmake /nologo /F Makefile.win32 clean
/bin/sh: line 0: cd: third-partyjudyJudy-1.0.3src: No such file or directory
fglock lunch & 14:43
agentzh ajs: you're on *nix?
ajs: thank you for your report! fixing 14:44
kolibrie agentzh: I'm planning on rebuilding pugs in a couple hours, and can test your Makefile
if no one else gets there first 14:45
agentzh kolibrie: that's fine.
kolibrie: ajs just found a bug in Makefile.PL
and i'm now fixing.
ajs Work thing coming up... brb
agentzh that's...really a stupid mistake. :P
ajs: see you! 14:46
ajs++
ajs Ok, taken care of. 14:47
Let me know when I should try again 14:48
agentzh okay!
ajs: now you can try again. :) 14:50
committed already. 14:51
svnbot6 r13445 | agentz++ | [Makefile.PL]
r13445 | agentz++ | - fixed a stupid OS detecting mistake. (thanks ajs++ for
r13445 | agentz++ | reporting this bug)
ajs Trying
agentzh it at least works for both my cygwin and windows. :) 14:53
pasteling "ajs" at 63.107.91.99 pasted "judy run failed" (80 lines, 3K) at sial.org/pbot/19794 14:54
agentzh oh... 14:55
looking now
ajs Looks like a Makefile wasn't regenerated that was needed in a "make clean" step 14:56
14:56 cjeris joined
agentzh weird. 14:56
it shouldn't have triggered in the perl Makefile.PL step. 14:57
14:57 buetow joined
ajs Sec, let me put in some echos between commands. I think Makefile.PL had just finished 14:58
Yeah, that "cd third-party..." is the first line of output from "make cleanjudy", and it gets no further than that, reporting that there is no rule to make target 'clean' 14:59
In the totally annoying gnuish `clean' quoting... grrr...
agentzh ajs: the error messages are from "make cleanjudy" rather than "perl Maekfile.PL", right? 15:00
ajs Yep
agentzh got it.
ajs Yeah, it looks like there is no Makefile in that directory at all right now
agentzh fixing
ajs: "make fast" and then "make cleanjudy" 15:04
i really don't want to use Makefile.win32 to clean judy on *nix. ;) 15:05
ajs is make fast an alias for make unoptimized? 15:06
agentzh it seems that your pugs is a fresh checkout.
aye
or "make optimized" if you prefer.
ajs agentzh: My pugs has been updated to version 11,000-something and back again, so it's going to be fairly cleaned out.
agentzh *nod* 15:07
ajs But, one would presume that cleanjudy should depend on fast if it needs it?
agentzh cleanjudy depends on the judy building not directly on "fast".
ajs make is still running 15:08
agentzh it requires some time. 15:09
ajs agentzh: I'm just saying, it should have a depend chain that allows it to work "out of the box"
agentzh ajs: okay, i'll try a fix then.
a sec...
fixed 15:11
committing now
ajs I'll let the first run finish, then I'll re-run after a make distclean to see what the new one does
svnbot6 r13446 | agentz++ | [Makefile.PL] 15:12
r13446 | agentz++ | - uses Judy's Makefile.win32 to do cleanjudy (a bug
r13446 | agentz++ | discovered by ajs++)
ajs Finished with no errors. Does judyclean just do a make clean in the judy dir?
agentzh ajs: don't make distclean.
that's not needed.
ajs: update first 15:13
ajs I was just going to make sure your dependency fix worked
agentzh use r13446
fine
ajs Re-running now
no distclean
agentzh k
for testing r13446, simply remove third-party/judy/Judy-1.0.3/src/Makefile will suffice. 15:14
*removing
and also run "make cleanjudy" multiple times... 15:15
15:15 discordja joined
agentzh sorry, i mean third-party/judy/Judy-1.0.3/Makefile 15:15
distclean won't remove Judy's Makefiles, i'm afraid. 15:16
even realclean won't either.
so one have to do that by hand.
*has
ajs agentzh: Hmm... that's not happy
agentzh which part is not happy? 15:17
ajs That distclean doesn't distclean
agentzh heh
ajs Multiple runs of cleanjudy works
However ... "perl -MExtUtils::Command -e rm_f JudyCommon\JudyMalloc.o" please tell me something is translating that backslash 15:18
agentzh oh well 15:19
does "\" work for you?
i'll convert \ to /.
fixing now
ajs afk
agentzh oh...there's no easy fix...
i think perl is able to handle that automatically, isn't it? 15:20
okay, Makefile.win32 is not *nix friendly...i'll revert r13446... 15:23
well, very few people will run "make realclean" or "make cleanjudy" on a fresh pugs checkout...(unless testers, for sure) 15:25
so we can safely ignore this corner case.
svnbot6 r13447 | agentz++ | [Makefile.PL] 15:30
r13447 | agentz++ | - reverted r13446, since Makefile.win32 is not *nix- 15:31
r13447 | agentz++ | friendly. so we will simply ignore the corner case
r13447 | agentz++ | that the user does "make realclean" and
r13447 | agentz++ | "make cleanjudy" on a fresh pugs checkout.
r13448 | audreyt++ | * Remove some redundant imports. 15:34
r13449 | audreyt++ | * "do {...} while 1" is now explicitly disallowed.
r13449 | audreyt++ | (same applies to other statement modifiers."
agentzh yay! audreyt++
so more tests should pass now in do.t. 15:35
agentzh thinks he needs a fast machine.
okay, win32 builds okay from scratch. 15:36
[particle] how can you run make anything after make realclean? doesn't realclean remove the makefile? 15:37
agentzh cygwin build is broken.
[particle]: it's Judy's Makefile, not pugs'
15:37 justatheory joined
[particle] makefile.pl doesn't create judy's makefile? 15:38
agentzh particle: it doesn't.
[particle]: for some reasons that i *do* know. :)
15:38 ivas joined
agentzh it's impossible to generate a Judy Makefile on Win32. 15:39
fglock anyone help me trace a perl5 problem? in v6, Test::report_failure() - Match objects are being stringified as "Pugs::Runtime::Match=REF(0x87dca7c)" - the overloaded stringification is not being called; it works elsewhere
agentzh but we can make it so on *nix.
15:39 miyagawa_ joined
[particle] agentzh: well, it could be as simple as a rename 15:39
agentzh [particle]: rename what?
[particle]: thank you for suggesting generating Makefile for Judy dureing Makefile.PL 15:40
[particle] np
agentzh i'm adding that now.
[particle] i don't understand makefile.pl much... i'm too used to the configure.pl in parrotland
agentzh (just for *nix users, since win32 users already have their own Makefile.win32)
[particle] so, that should remove the corner case?
agentzh yep! 15:41
[particle] yay!
agentzh :)
15:41 gunya joined
agentzh oh, the building logic for *nix judy is bogus. 15:43
at least it seems to be.
i won't fix that since i can't test it locally.
is mac's shell similar to *nix? 15:45
[particle] yes, mac's os is bsd-based iirc 15:46
agentzh okay 15:47
people on mac is lucky. 15:48
*are
particle: Judy's configure takes quite a long time...
[particle] i see, so rather than subject windows users to a slow process, the result is included instead? 15:49
agentzh i'll check the existence of Makefile and config.status before configure.
[particle]: sort of. 15:50
:)
win32's judy building is *damn* fast. :)
fglock mm - just adding this line makes the program die: if (overload::Overloaded $got) { print "Overloaded\n" } - but it only dies several statements later 15:51
svnbot6 r13450 | agentz++ | [Makefile.PL] 15:52
r13450 | agentz++ | - generated Judy's Makefile during the "perl Makefile.PL"
r13450 | agentz++ | phase as suggested by [particle]++ (this patch only
r13450 | agentz++ | affects non-Win32 users.)
agentzh anyone on *nix willing to test r13450 for me? 15:53
agentzh thinks it is a long day for him.
svn log ~= s/generated/generates/ 15:57
ajs svn log ~~ s/ generated / generates / 15:58
mauke parse error before =
agentzh :P
that is p5. :)
mauke no, =~ is
ajs How many people do you think will make the mistake of thinking that whitespace in the substitution spec is ignored?
agentzh oh, sorry... 15:59
ajs: have some cycles to try out r13450?
ajs Yeah, ~= in P6 is going to freak me out for a while
agentzh: sure
agentzh it should have fixed all your problems.
thanks to [particle]++'s suggestion. 16:00
ajs All my problems? What about that burning sensation?
clean run, thanks
agentzh Judy's Makefile is now generated at "perl Makefile.PL"
ajs nice
agentzh so "make cleanjudy" on a fresh checkout is fine now.
agentzh has been making so many stupid mistakes since he has well passed his bedtime. 16:04
my brain gets fried... 16:05
[particle] "this is your brain on pugs" 16:06
agentzh (?) 16:07
kolibrie agentzh: I just finished my svk pull, which commands do you want tested?
agentzh great 16:08
"perl Makefile.PL"
tell me if you see the Judy configuration process.
and also the rev number
sorry, ignore the "rev number" crapt. 16:09
kolibrie it did say Configuring Judy...
and then a bunch of automake-style checks 16:10
[particle] agentzh: en.wikipedia.org/wiki/This_Is_Your_...n_On_Drugs
agentzh what's right below that line?
lambdabot tinyurl.com/jf9cx
agentzh [particle]: looking
kolibrie: oh, that's right. 16:11
kolibrie agentzh: checking whether to enable maintainer-specific portions of Makefiles... no 16:12
agentzh cool
kolibrie: that is correct behavior.
kolibrie agentzh: anything else before I 'make'
agentzh "make judyclean" 16:13
sorry, i mean "make cleanjudy"
kolibrie done
agentzh any errors/ 16:14
?
so there's no error at all?
kolibrie just recursed through a bunch of directories, and all looks okay 16:15
agentzh cool
kolibrie now make?
agentzh check out third-party\judy\Judy-1.0.3\src to see if any subdirectories contain .o files
and .c files which are not under version control. 16:16
kolibrie tries to do the backslash to forward-slash conversion in his brain
agentzh kolibrie: sorry, i copied that directly from my win32 environment.
kolibrie no .o files, only on .c file: Judy.h.check.c 16:17
s/on /one /
agentzh kolibrie: that .c file is under version control, right?
how about subdirectories JudyL and Judy1?
are there any .o files? 16:18
JudyL and Judy1 should not hold any .c files at all.
kolibrie JudyL and Judy1 have no .o nor .c files 16:19
agentzh yay!
so now do "make fast"
that is the last thing to test.
kolibrie quickly makes 16:20
agentzh tell me the rev number displayed 16:21
kolibrie at the end, or somewhere in the output?
agentzh somewhere near the beginning 16:22
something like "Current version is 13436"
if you have missed that, then ingore that. 16:23
kolibrie Current version is 13450
agentzh thanks!
that is what i expected.
and the build is going smoothly?
svnbot6 r13451 | audreyt++ | * unTODO one passing test.
kolibrie so far so good
agentzh excellent 16:24
oh, audreyt is still in hacking mode.
0:30 AM for her.
(for me too)
16:24 zgh joined
kolibrie only hacks between 05:00 and 20:00 16:25
agentzh kolibrie: have you seen the line "Compiling Pugs.Version"?
kolibrie haven't been watching
agentzh somewhere in the middle... 16:26
kolibrie it's on pugs-fps-0.7 right now, would that line be above that, or still to come
agentzh oh, below it!
you're still building prerequisites 16:27
kolibrie is running on a slow machine today
agentzh :)
agentzh has caught a bad cold and he is not very conscious atm. 16:29
kolibrie Configuring Pugs-6.2.12... 16:30
agentzh good
a short way to go.
kolibrie: you're building from scratch, right? 16:31
a typical "make fast" should take so long.
s/should/should NOT/
kolibrie agentzh: last build on this computer was at least a month ago 16:32
agentzh anyway, it's a more thorough test though much more time is required. :)
kolibrie++
kolibrie Compiling Pugs.Version ( src/Pugs/Version.hs, dist/build/Pugs/Version.o ) 16:33
agentzh right, that's it. :D
kolibrie glad I could help 16:34
agentzh so once you have built pugs, the test is successful completely. :) 16:35
kolibrie: thanks for your feedback!
kolibrie shucks, GHC's heap exhausted 16:36
agentzh kolibrie: after the build, run "make cleanjudy", and see if the .o and .c files are deteleted as expected.
kolibrie: simply rerun "make fast" from that point.
kolibrie thought so 16:37
fglock I'll need to workaround the overload error - I can't find any references to it in google
agentzh fglock: good luck! (although i can't help) 16:38
kolibrie fglock: I can maybe test something for you here after my build finishes
fglock kolibrie: I created a 'nomethod' to trace the problem - the stringification function is not being called at all 16:40
kolibrie fglock: anything I can help you with? /me really doesn't understand what fglock is battling 16:42
16:43 Psyche^ joined
agentzh well, i can't stay up any more...kolibrie: simply report errors (if any) here. i'll backlog tomorrow. 16:44
fglock kolibrie: do you have v6.pm installed? try running perl5/Pugs-Compiler-Perl6$ perl -It ../../t/xx-uncategorized/rule_recursive.t
kolibrie agentzh: ok
agentzh g'night everyone & 16:45
agentzh waves
16:45 agentzh left
fglock kolibrie: you will see errors like " Actual: 'Pugs::Runtime::Match=REF(0x8836228)' " 16:45
because the Match object is not being stringified
kolibrie fglock: will try in a bit when my build is finished 16:47
(all CPU and memory are being used)
16:51 frederico joined
fglock clkao: ping 16:54
clkao fglock: pong 16:55
fglock clkao: any chance that Data::Bind changes overloading somehow? 16:56
clkao overloading? 16:57
i don't recall overloading in data::bind
fglock clkao: did you see the error I mentioned above? 16:58
clkao no, looking
which one?
16:59 Psyche^ is now known as Patterner
fglock clkao: when you pass a Match object in '$got' to Test::report_failure(), the object can no longer stringify using overload 17:00
but overload::Overloaded($got) reports that it is still overloaded
clkao uhm. make a small test case so i can look at after dinner? 17:01
fglock ok
kolibrie agentzh: make cleanjudy left no .o files in Judy-1.0.3/src, and just that one .c file, but JudyL and Judy1 have lots of .c files 17:06
[particle] kolibrie: i think it's only important if those .c files *are not* in the repo (eg if they're generated) 17:07
17:09 lampus joined
TreyHarris ajs: i wasn't earnestly making either assertion, i was just trying to figure out what ruoso was talking about and trying to provoke him to elaborate on his thoughts. 17:10
lambdabot TreyHarris: You have 1 new message. '/msg lambdabot @messages' to read it. 17:11
svnbot6 r13452 | agentz++ | [third-party/judy/Judy-1.0.3/src/Makefile.win32]
r13452 | agentz++ | - commented out CC= and COPT= since that will possibly
r13452 | agentz++ | break other windows users' judy build.
r13452 | agentz++ | ([particle], Limbic_Region: please check if it works for
r13452 | agentz++ | you. :))
TreyHarris @messages
lambdabot agentzh said 2h 49m 52s ago: the pugs -v rev number problem ought to have been solved.
kolibrie [particle]: they were not there before I did my make
17:11 kanru joined 17:13 lanny joined
pasteling "fglock" at 200.17.89.34 pasted "possible overloading problem with Data::Bind" (27 lines, 728B) at sial.org/pbot/19799 17:16
lanny So here's a question. I'm working on writing sort() in Perl 6. My tests of that sort are unexpectedly succeeding because they are using the builtin sort(). Is there a way to turn off the builtin so my tests successfully fail? 17:18
pasteling "fglock" at 200.17.89.34 pasted "fixed test case - was "possible overloading problem with Data::Bind"" (27 lines, 729B) at sial.org/pbot/19800
fglock lanny: add a package name? like My::sort 17:19
[particle] lanny: you're writing in what language? 17:20
lanny Yeah. Thought of naming it msort() (based on merge sort).
I'm writing it in Perl 6.
fglock lanny or use it as a method: $self->sort()
[particle] and it's using the builtin perl6 sort?
fglock $self.sort then
lanny particle: yes. lots of changes since I last commmited but t/unspecced/sort.t
[particle] why not override the sort op? 17:21
lanny The way it's defined in the synposis it's not an op.
fglock clkao: sial.org/pbot/19800 shows the problem
lambdabot Title: Paste #19800 from "fglock" at 200.17.89.34
17:21 mago joined
Limbic_Region what am I supposed to check out if it works for me? 17:21
17:22 lisppaste3 joined
clkao fglock: oki 17:22
fglock Limbic_Region: are you talking to me?
Limbic_Region no
lanny fglock: Thanks. That would probably work given the Array:: and List:: definitions (i.e., $thing->sort()).
Limbic_Region [13:11] <svnbot6> r13452 | agentz++ | ([particle], Limbic_Region: please check if it works for you 17:23
that's what I am talking about fglock
agentzh ping
ajs lumi: thanks for helping me chase down that @*CHECK bug in -BPIR, but of course, it turns out to be only the FIRST of the errors. -BPIR is simply quite broken. 17:38
17:48 integral is now known as ntgrl, integral joined 17:50 lisppaste3 joined 17:53 glasser joined
svnbot6 r13453 | ajs++ | Add in @*CHECK for PIR backend. Can someone please verify this is correct? 17:53
17:57 SubStack joined 17:58 ofer0 joined
pasteling "ajs" at 63.107.91.99 pasted "r13453 regex error that wasn't happening yesterday" (29 lines, 964B) at sial.org/pbot/19802 18:01
ajs Anyone else seeing that/
?
Oh freaky. I'm getting different results each time I run my regex example under v6 18:08
fglock ajs: looking 18:10
[particle] v6 builds up compiled mods as it goes, so the first few times may vary
(at least i've seen this before)
ajs [particle]: The main difference seemed to be between when I ran v6 from inside my pugs tree and when I ran it from outside, but then cding back into pugs yielded no change back 18:13
18:18 fglock joined
fglock ajs: try this (works for me): perl -e 'use v6-alpha' - ' regex b { 1 }; "1" ~~ /<b>/; $/.perl.say; ' 18:19
ajs Trying it 18:22
still get: Can't locate object method "b" via package "Pugs::Grammar::Base" at (eval 121) line 7. 18:23
fglock ajs: there was a fix for this error yesterday, maybe reinstall v6 ? 18:24
ajs did 18:25
fglock it was "r13430 | fglock++ | v6.pm - added headers to allow rule definitions outside of grammars" 18:26
perlbot What kind of idiot karmas himself? Your kind of idiot!
ajs Ah... ok, how do I install v6 from the tree instead of CPAN? Do I just CD into the perl5 subdirs and run make install? README says that's bad 18:28
fglock ajs: that should work
18:34 bernhard joined 18:37 buubot joined 18:40 mdiep joined 18:45 larsen joined
svnbot6 r13454 | lanny++ | [t/unspecced/sort.t] 18:47
r13454 | lanny++ | - cleanup to better handle @by with traits
r13454 | lanny++ | - tests for most examples in www.nntp.perl.org/group/perl.perl6....uage/16578
lambdabot Title: nntp.perl.org - perl.perl6.language (16578), tinyurl.com/j9pt5
svnbot6 r13455 | fglock++ | Pugs-Compiler-Rule - the Rule grammar now allows array and hash aliasing 19:02
r13456 | fglock++ | Pugs-Compiler-Rule - comment out debugging info
19:02 dakkar joined 19:03 buubot joined 19:04 dakkar joined 19:12 KingDiamond joined 19:20 nferraz joined 19:25 polettix joined
ajs fglock: README in perl5 should be updated with two things: 1) it's ok to install this stuff and 2) the order in which it should be installed. A Makefile would be awesome, but I'll settle for a README 19:30
19:32 lkjasa joined, dduncan joined, lkjasa left 19:33 theorb joined
dduncan agentzh, fyi, I did a pull today, and a 'make' ... and the version is being updated, it seems ... make also rebuilt all of the Haskell 19:33
lambdabot dduncan: You have 1 new message. '/msg lambdabot @messages' to read it.
dduncan hm, I'm going to have to learn how to leave messages on this system myself 19:34
I saw the backlog before coming here, fyi
[particle] @tell dduncan remember to use @tell <user> <message> to leave messages yourself 19:36
lambdabot Consider it noted.
[particle] scnr
dduncan whaddayagot?
lambdabot dduncan: You have 1 new message. '/msg lambdabot @messages' to read it.
dduncan thatwhatyagot!
Limbic_Region and if you don't want the entire channel to know what you are telling <user> use: /msg lambdabot @tell <user> <msg> 19:37
dduncan figures
[particle] oh, i'm so glad you can msg lambtabot. it is a bit noisy.
TimToady of course, that doesn't help if they just say:
lambdabot TimToady: You have 1 new message. '/msg lambdabot @messages' to read it.
TimToady @messages
lambdabot agentzh said 5h 15m 32s ago: the pugs -v rev number problem ought to have been solved.
ajs fglock: Thanks for the help, updating to most recent v6 from the pugs tree helped
dduncan I knew before that you could use /msg syntax to private message anyone
fglock ajs: np 19:38
dduncan come to think of it ...
PerlJam lambdabot needs to be quieter. Surely it can /msg people?
19:39 larsen joined
dduncan hm, I tried to /msg perlbot but it didn't work 19:39
useful to try out things without spammin the list
pasteling "ajs" at 63.107.91.99 pasted "build.sh script I ended up using in perl5 in case anyone cares" (11 lines, 320B) at sial.org/pbot/19805
TimToady I wonder what a build system in P6 would look like... 19:41
ajs use Build; my Build $doit .= new :universe<.>; 19:42
;)
TimToady your formulation seems to be missing the dependency info... 19:43
I'm trying not to use the "m" word here... 19:44
or the "nm" word, or the "gm" word...
or the "pm" word, for that matter.
[particle] @sources>>.build(@args);
lambdabot Unknown command, try @list
dduncan yes, that's what it would look like 19:48
svnbot6 r13457 | fglock++ | Pugs-Compiler-Rule, Pugs-Compiler-Perl6 - updated README files - ajs++ 19:50
dduncan question: I understand that a number of large changes will go into the trunk or be made active following 6.2.13, for inclusion in 6.28.0, but what is the nature of those changes? ... 19:52
svnbot6 r13458 | fglock++ | v6 - update usage (pod)
dduncan eg, will there be some exposure of the metamodel to perl 6 code?
TimToady the metamodel is available via .HOW 19:53
dduncan oh, and with r13454, 'make' does update the reported version correctly
thank you agentzh
TimToady yes, but the problem with 'make' is that it's about 582 different programs... 19:54
dduncan I will now try another pull and see if it keeps working 19:55
regarding .HOW, yes that exists, but a bunch of useful functionality isn't there yet; eg, the likes of .attributes doesn't seem to be functional yet that I could see 19:56
so I wondered if fleshing this out was planned to come soon, or if it is being put off for awhile behind other things
TimToady that is the next step after this release is out, as far as I know. 19:57
19:57 nferraz left
dduncan great! 19:58
[particle] also i believe, dropping support for ghc 6.4
TimToady *sniffle*
dduncan and I apologize if I have said anything that seems insensitive
over the last few days 19:59
TimToady and if you didn't, you don't? :)
I have not perceived any perceived problem.
20:00 larsen joined
dduncan I just said that because I didn't know if I unintentionally said anything that sounds too demanding ... while I never intend to make people angry, it can happen 20:00
oh, and the version stuff seems to be continuing to work! 20:01
my pugs -v reports: r13458
with a simple pull and make
on another matter, TimToady, thanks for the handwaving you added in [svn:perl6-synopsis] r12185 - that helps clear some things up 20:04
TimToady well, it's still full of holes... 20:05
dduncan re class naming
TimToady but what we're aiming for is something that's really neither name equiv nor struct equiv.
dduncan yes, but it brings out the idea that parameters can be taken into account when defining when a class is comparable to another
indeed 20:06
TimToady it's more like, are these provably generated from the same immutable raw materials?
if so, they're the same.
20:06 lisppaste3 joined
TimToady which in a way is closer to structural typing. 20:06
but it's not duck typing, because that's accidental name collisions from differently generated types 20:07
dduncan well, I know what I was trying to achieve with my related request, but it is difficult for me to express or simply to know about the best way to achieve it
TimToady so maybe we need a new term, generative equivalence, or some such. 20:08
20:09 integral is now known as ntgrl, integral joined
dduncan fundamentally, the structures I'm trying to define are immutable and have immutable components, so it should fit into the immutable raw materials requirement 20:10
they are just restricted Mapping, really
though it would be nice to syntactically use the elements like class attributes
TimToady roles may be sufficiently immutable 20:11
dduncan yes
TimToady well, speaking of mutating, I need to mixin the Job role. 20:14
dduncan indeed
TimToady commutate &
dduncan good work day 20:15
TimToady & etatummoc 20:29
[particle] how symmetrical 20:32
fglock is there a way to make inheritance work in $obj->$method (p5) 20:39
svnbot6 r13459 | fglock++ | Pugs-Compiler-Rule - implemented Matching against non-strings
r13459 | fglock++ | my $rule = Pugs::Compiler::Regex->compile( '<.meth("param")>' );
r13459 | fglock++ | my $match = $rule->match( $object );
clkao fglock: i can reproduce 20:47
fglock clkao: good 20:48
20:49 jferrero joined 20:58 larsen joined
svnbot6 r13460 | fglock++ | Pugs-Compiler-Rule - implemented <at($pos)> 21:01
21:03 vel joined
fglock tomorrow is holiday ! 21:09
21:11 Limbic_Region joined
dakkar fglock: whre? 21:11
fglock dakkar: yes
21:11 SubStack joined
dakkar ehm... that was a "where?" (my mistype) 21:12
fglock ah, where - Brazil :)
dakkar you see, I'm in the US at the moment, but I don't know the local holidays ;-) an unplanned day out of work would be gladly received ;-)
fglock sure - I just found out :) 21:13
21:14 shachaf_ joined 21:18 foo\ joined 21:21 BooK joined
svnbot6 r13461 | fglock++ | Pugs-Compiler-Rule - fixed method dispatch problem while matching against non-strings 21:24
21:28 larsen joined 21:32 mdiep_ joined 21:39 fglock left 21:43 DaGo joined 21:46 larsen joined 21:54 dduncan left
svnbot6 r13462 | clkao++ | Make arg_bind work with overloaded objects. 21:58
22:01 mago joined 22:02 jdv79 joined 22:14 eden_c joined 22:18 wilx joined 22:34 dduncan joined
dduncan question: are we using "pass" for anything in Perl 6 right now (ignore the test suite) 22:34
if appropriate, that's my new suggestion for a renamed "grep" 22:35
as already posted to p6l
(some rationale given there)
22:37 dduncan left
svnbot6 r13463 | putter++ | Common/backtrack_api.pl created. 22:37
r13463 | putter++ | 2006-09-19 - The idea here is Perl6::Compiler::Rule needs ranges
r13463 | putter++ | {n..m} and tailcalls implemented, and perhaps performance tuning.
r13463 | putter++ | The old regexp_spike code has these. This file is a snapshot of
r13463 | putter++ | a quick attempt to strip the old code down to be easily used by
r13463 | putter++ | perl5/Pugs-Compiler-Rule/lib/Pugs/Emitter/Rule/Perl5,pm
22:37 Aankhen`` joined 22:38 ludan joined
ludan re 22:39
22:40 putter joined
putter fglock: ping? 22:42
lambdabot putter: You have 1 new message. '/msg lambdabot @messages' to read it.
nothingmuch hi putter
just in time ;-)
fwiw MO is just as comitter friendly as pugs except i don't need to know your email to give you a commit bit 22:43
avar Hmm, mm
nothingmuch you just need to have one ;-)
putter hi nm. just sent you a :), but wrong channel. :/
avar It would be neat if Perl6 included a bug report module to make ${yourapp}bug programs 22:44
nothingmuch avar: i'd say that is definitely premature project management ;-)
putter re commit friendly, great. likely wont happen before the weekend though.
avar Not at all, I think it would be neat in any language stdlib, nothing to do with Perl 6 as such
Well maybe not any stdlib, not C's for instance: 22:45
nothingmuch avar: it *does* have to do with whatever the ExtUtils::MakeMaker replacement will be
and it *should* be a plugin
avar nothingmuch: *In**deed*
nothingmuch is just making fun of RFC language =)
fwiw, all it really needs is to send mail 22:46
TimToady to MicroSoft.
avar What's the best way to get into writing some of that stuff?
nothingmuch write a module to abstract SMTP and Sendmail with the same API
and write a wrapper that will try to send mail very hard, by using either this or that
avar nothingmuch: which subset of the api?
nothingmuch and the MX host from the target mail address unless it can learn who the host is from the system 22:47
neither
avar You want SMTP along with milter callbacks?
nothingmuch just $sender->sendit( $RFC2822 )
ick no
look at markov's stuff
it's slightly overengineered (too much effort, too much magic)
but in general it's a good start 22:48
avar Where's his stuff?
nothingmuch once you have that you just need a template and a command line module thingy
start with Mail::Box
and also look at Email::Simple
(which is arguably a little too simple)
and also talk with rjbs
(on irc.perl.org)
avar I don't want to do mail yet:)
nothingmuch he should be able to provide lots of insight on what needs to be done better for the p6 mail stuff
then what do you want?
avar For the purpose of this question I want to write Perl 6 modules, is it best to just commit them to pugs? 22:49
nothingmuch oh
yes =)
22:49 eden_c left
nothingmuch do you have a commit bit? 22:49
avar yeah:) 22:50
nothingmuch okay, all set then =)
ext/
putter @tell fglock so the idea on backtrack_api is it's a helper class providing code strings which abstract the combining of backtrackting subs taking one (continuation) argument, and keep application state in local()s. local()s tend to be faster than passing as args. once in a form which requires little change from Emitter::Perl5.pm, we can try it and see. Has ranges and tailcalls, and is well tested.
lambdabot Consider it noted.
avar But is there anything particular I should read besides the obvious? Like good examples of porting (and bad ones)?
nothingmuch @tell lambdabot it's a cool bot
lambdabot Nice try ;)
nothingmuch @botsnack 22:51
lambdabot :)
nothingmuch avar: there's a porting FAQ
and... err... just ask around for feedback
once you have code people can comment
avar yay:)
avar goes crazy
nothingmuch the best thing to do is to not be shy
commit often, commit early
^_^
anymoose, sleepy time 22:52
putter g'night &
23:06 justatheory joined 23:26 SubStack joined 23:28 iblechbot joined 23:30 cjeris left 23:56 knewt joined 23:59 SubStack joined

Powered by Cro and the Raku® Programming Language.

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

Thank you!