04:03 sorear joined 05:24 sorear joined 12:51 masak joined 18:10 mberends joined 18:53 masak joined 18:57 diakopter joined
diakopter lurks as usual 18:57
masak o/ 19:00
spinclad \\o
moritz_ o|
jnthn o/
colomon \\o 19:01
jnthn pmichaud: #rs!
moritz_ any pmichauds around?
pmichaud here.
jnthn oh wow, we almost has, like, good attendance \\o/
colomon \\o/
moritz_ bkeeler? 19:02
sorear?
shall we start with masak's gsoc report?
colomon :)
jnthn Why not. :-)
masak clears throat 19:03
spinclad q1q
masak I've created a branch.
in that branch, I've created a few roles.
the preexisting test file now runs.
not all tests pass. :)
I had a question for TimToady about .chars, which TimToady answered today.
moritz_ what's the immmediate plan? any blockers? 19:04
masak that's it so far. progress is moderate but good. this is officially Day 2.
immediate plan is to write more tests, and pass them.
jnthn Passing some new tests by day 2 is kinda good. :-)
masak: You planning to merge stuff to trunk fairly often, or after some review first, or?
masak: I don't want to end up with a branch with like 3 weeks worth of stuff to review, if we can avoid that. 19:05
masak jnthn: I'll keep that in mind.
jnthn OK, thanks.
masak actually, I haven't felt the need to be in a branch at all yet. but it's only been two days. :) 19:06
jnthn has seen too many Parrot GSoC projects in the past that never left their GSoC-long branch
masak yeah, let's get into the habit of merging regularly. :)
jnthn +1
colomon +1
pmichaud +1
masak .eor
jnthn OK. Who next? pmichaud?
pmichaud I reviewed more of sorear++'s REPL patch and brought a few very nice improvements into the nqp-rx mainline 19:07
we now have autoprinting of results from the interactive REPL, with Coke++'s help
moritz_ \\o/ 19:08
pmichaud I also reviewed bkeeler++'s regex interpolation code
bkeeler largely did what I was looking for, but after seeing the code I decided a different (and much less invasive) approach was called for
so, I refactored nqp and rakudo to use the newer approach
as a bonus, array-alternations of regex is now somewhat LTM-aware 19:09
masak what's the refactor? in twenty words or less :)
pmichaud bkeeler's version created a new PAST::Regex node type for interpolation
after re-reading S05, it occured to me that interpolation is really a special case of a subrule
so I rewrote it that way 19:10
another bonus is that things like <abc=@xyz> now work
so that you can capture the result of the alternated match
and we also get stuff like <!@xyz>
jnthn I guess <$xyz> and <@xyz> alone don't create xyz named captures automatically?
pmichaud they do not.
jnthn OK
jnthn is happy about that 19:11
moritz_ pmichaud: re autoprinting, did you notice that rakudo segfaults when an exception is thrown and not caught?
pmichaud moritz_: No, I didn't notice that. Should be relatively easy to fix.
jnthn pmichaud: More REPL work coming up next? 19:12
pmichaud anyway, merged the new interpolated regexes implementation
I'm currently working on REPL, yes
jnthn Excellent.
masak \\o/
pmichaud should have it ready by the end of the day.
jnthn is *so* happy to see pmichaud++ committing lots of stuff again
pmichaud tomorrow I plan to fix up closures, and the contextual PAST node type 19:13
spinclad \\o/
colomon +1
pmichaud after that is lists/arrays again, probably Friday-ish.
.eor
jnthn pmichaud: Is the parsing of some typenames in nqp-rx at all related to the grand plan surroudning contextual node? 19:14
pmichaud jnthn: no.
jnthn pmichaud: OK. I found it curious to see the ability to parse them being put in without any underlying meaning, that's all.
pmichaud it's more so we can look at supporting simple multi-subs
jnthn Ah, OK.
+1
pmichaud and potentially things like "my int $x", but I'm still not at all comfortable with that declaratory syntax. 19:15
jnthn In what sense?
The syntax itself, or the use of it to map to underlying Parrot register types?
pmichaud in Parrot, registers coerce, not constraint.
*constrain
jnthn Yes, that is true. 19:16
pmichaud so, the Perl 6 semantic of "my int $x" is that only int can be stored there.
jnthn *nod*
It'd be difficult to enforce that in nqp.
And probably not desirable.
pmichaud whereas the Parrot semantic of "my int $x" would be to convert anything bound to it to an int.
jnthn Well, maybe.
Aye.
Anyway, the multi thing sounds OK.
pmichaud ....and that's a bit too much of a semantic gap for me to be wanting to accept right at this moment.
jnthn Yeah, agree. 19:17
pmichaud so it may need to be "my int $x as int" or something like that.
jnthn *nod*
moritz_: Has you a report? :-)
TimToady or my int(Any) $x in the other proposal I made once
jnthn TimToady: Yeah, I didn't dislike that syntax.
moritz_ jnthn: moreorless 19:18
TimToady where Any was default, so int() would be coercive
moritz_ I've worked on various bits and pieces
one of them being match object
had some small success in the mob5 branch
a named, non-quantified capture now works with pure Perl 6 match objects
however everything arrayish (quantified or numbered captures) goes awefully wrong 19:19
null PMC access in Seq!fill
somehow Perl 6 arrays seem not to like it when things are being pushed onto them via vtables, or so
haven't figured out what exactly is wrong
pmichaud moritz_: it's all likely to be majorly refactored with array/list changes anyway
I'd be very reluctant to build lots of code on the existing array/hash implementation. 19:20
moritz_ pmichaud: problem is, match objects are kinda major blocker for me
they block JSON::Tiny and thus the book chapter on grammars
pmichaud I might be able to fix that more quickly.
we can discuss on #perl6
moritz_ apart from that I've done small things here and there 19:21
started a local branch to enable enums in the setting
hit an iterator issue I haven't figured out yet - still have hope to fix it on my own
applied small patches, reviewed advent integration tests
trying to lure more contributors - ideas for new challenges are very welcome 19:22
way too many ideas for not-so-small projects, too few tuits
.eor
jnthn moritz_++
colomon?
colomon moritz_++
Since last time, most of my Rakudo-effort went to the release. 19:23
Other than that, a few minor Numeric things.
Not sure how much connectivity I'll have over the extra long weekend coming up.
PerlJam greetings. 19:24
colomon o/
PerlJam finally shows up during a #rs 19:25
jnthn :-)
colomon: ero?
*eor?
colomon jnthn: I can't think of anything else to say, so yes. :)
.eor
jnthn :-) 19:26
For me, I had a pretty unproductive week last week...was oddly tired. Feeling better this one. :-) Got a few small Rakudo patches in that fixed some bugs.
And also did some work on book.
Got another book patch partially there.
Did some roadmap updates to try and keep it in sync with what's done in reality. 19:27
This week, plan to focus on book + bugs.
masak \\o/
colomon what's the deadline on the book?
jnthn .eor 19:28
moritz_ colomon: too early :(
jnthn Soon. :(
moritz_ colomon: we don't have a hard one, I think
jnthn Anyone else have some report, btw?
moritz_ but after our deadline it will take a few weeks for it to actually be in print, or so
[particle] the original intent was hardcopies at yapc::na
colomon ah, I thought chromatic's publishing needs meant the deadline was next week.
PerlJam didn't chromatic say in one of his emails that he needed 3-4 weeks lead time or something?
pmichaud we also expected to have Rakudo * out by now, though. 19:29
jnthn That too.
moritz_ so what's the current schedule for R*?
[particle] make it good, then make it right, then make it on time
jnthn If we're still shooting for June - which I believe we are - that means we have ~ a month to go.
pmichaud shortly after June release.
moritz_ so, "late in June"?
pmichaud It can't be before June Parrot release 19:30
so.... around the June release date
moritz_ June parrot release is 15th
jnthn It also needs to build against the June Parrot release.
moritz_ which is as early as it can be
so... shall we say around the 22nd?
pmichaud yapc::na is jun 21-23, so around then might be good.
jnthn So essentially, any improvements that need nqp-rx hackage will need to go in before then.
What are the Real Big Things that must happen for release? 19:32
colomon We have four weeks, then.
jnthn We have the ROADMAP must-haves, of course.
colomon no, wait, three?
moritz_ has blogged a list
pmichaud four, at least
colomon four.
exactly if we're talking the 22nd.
jnthn I also think we need to sort out closures, match objects, arrays/lists somewhat, along with the ROADMAP thngs.
pmichaud five if we're talking the 22nd
jnthn *things
pmichaud oh wait, four. 19:33
silly calendars.
I think that closures and arrays/lists are absolutes, which is why they're high on my plate
REPL was another absolute, and I'm doing it now.
jnthn perlgeek.de/blog-en/perl-6/ for moritz_++'s list, btw 19:34
pmichaud some form of Whatever handling would seem very important
anyway, by next week's #rs I'll report on must-haves. 19:35
jnthn Backtraces with Perl 6 file names and line numbers - I can take this.
I planned to do it last week and lacked the energy.
pmichaud I have energy this week, and there are a few times upcoming that are bad for coding but good for writing lists.
jnthn +1
PerlJam what is "installation standards" under Ought-to-haves? Is that directory layout and such for Rakudo?
pmichaud but anyone feel free to pre-empt me if you get the tuits :)
jnthn pmichaud: On "/<foo>/ should pick up a regex named foo from the lexical scope, not only from method lookup" 19:36
moritz_ we kinda have that, vaguely
jnthn pmichaud: How hard is that to do?
pmichaud: I did the stuff so my regex foo { ... } works
TimToady the easiest way to *op* might be with a rewrite rule in the grammar that knows the exceptions like ranges
that would be close to spec anyway
pmichaud jnthn: a lexical would always override a method, then? 19:37
jnthn TimToady: I was pondering that, and the "is this an exception" hook would just know the exceptions hard-coded, and we could fill it out later.
TimToady that would be fine for R*
jnthn TimToady: What pmichaud asked ;-)
pmichaud: <.foo> is always method and <&foo> is always lexical, though, iirc. 19:38
pmichaud in some sense I'd prefer to see <&foo> always be used for the lexical.
jnthn And <foo> would "dwim"
pmichaud not the least of which is because I can implement that in about two keystrokes.
TimToady tempting 19:39
jnthn pmichaud: If we implement <&foo> we will mitigate a lot of people's immediate problem.
pmichaud (<foo> "dwim") I'm not always sure that we know "wym" though.
spinclad (but <.foo> is don't-record-foo?)
TimToady anything with punctuation on the front is don't-record
pmichaud anyway, having <foo> automatically find things in the lexical scope is a bit tricky, because the method name doesn't have a & and the lexical name does. 19:40
jnthn <foo=&foo> # records
spinclad ah. so <.foo> can be method
pmichaud technically, <.foo> *is* method :-)
jnthn Or <foo=.foo>
TimToady it already is
spinclad :)
pmichaud anyway, if someone can let me know if <foo> is lexical priority or method priority, I can add it. 19:41
TimToady requiring & seems conservative, though if we change the dwimminess of <foo> it could break things later
seems fairly unlikely though
masak .oO( when syntaxes clash )
pmichaud right now <foo> only looks at methods in nqp-rx
jnthn masak: From a user perspective, is having to write <&foo> for a "my regex blah { ... }" outside of a grammar going to be a pita? 19:42
(iyo, of course :-))
TimToady well, I think it has to be lexical priority if it dwims, since we know that at compile time, and not the other
pmichaud anyway, having <&foo> work is a short-term "I can do this now" sort of fix.
moritz_ it would be a large improvement 19:43
pmichaud being able to make <foo> dwimmy on lexicals is a bit more work, but potentially do-able.
moritz_ even if people still expect /<foo>/ to pick up from lexical scope
pmichaud (I agree with TimToady++ -- any dwimminess needs to be towards lexicals.)
jnthn pmichaud: Does that we already know all the lexicals at compile time help us?
pmichaud jnthn: yes. 19:44
jnthn pmichaud: That is, could we code-gen <foo> as if it were <foo=&foo>?
19:44 masak joined
TimToady yes 19:44
jnthn (now easily I mean)
pmichaud jnthn: yes, that's what would happen.
jnthn OK, excellent.
pmichaud The PAST::Regex node already knows how to handle aliasing.
anyway, I'll commit to <&foo> immediately, and may have <foo> shortly thereafter. 19:45
jnthn pmichaud++
moritz_: I'll try and get to "Currently user-defined operators hide candidates of the same name from outer scopes. This is rather annoying." in the next week too.
moritz_ \\o/
jnthn I'll make that and backtraces my main goals. :-)
And book :-) 19:46
TimToady and everything else :-)
jnthn TimToady: That can wait for the week after. :-P
OK, any more things for this week? 19:47
spinclad: I think you q1q'd?
spinclad as pure administrivia, can we get someone op'ed so we could set a /topic? (for meeting time, eg)
pmichaud can someone write some <&foo> tests?
moritz_ pmichaud: I can do that tonight (ie in the next 1 hour)
pmichaud moritz_: that would be a huge help.
jnthn moritz_++, pmichaud++
moritz_ so, <&foo> does not capture, right?
TimToady but then I'll be off topic :(
correct 19:48
pmichaud moritz_: it does not capture.
TimToady nothing with punctuation captures
pmichaud moritz_: but you can do <foo=&foo> to capture.
moritz_ great
sure think
s/k/g/
TimToady you don't have to remember which characters capture, other than alphanums
pmichaud right... any punctuation means "no capture"
19:49 masak joined
spinclad i think channel creator has op grant privilege? 19:50
jnthn I'm...not sure how created it. :-)
spinclad or talk to freenode mgmt
pmichaud I've not had much luck with freenode mgmt, fwiw.
Tene It's not registered. 19:51
moritz_ easiest solution: everybody leave
Tene Yeah, that's about the only reasonable option.
pmichaud maybe one of us could just kic..... oh.
19:51 masak left
Tene or choose a new channel name, and actually register it when you create it. 19:52
pmichaud #rsketch
#rakudoreport
#rakrep
moritz_ #rs
Tene #rakudorakudorakudo
moritz_ that's how we call it anyway :-)
pmichaud there's already a #rs
moritz_ oh wait, taken :( 19:53
TimToady radioshack?
PerlJam #bikeshed ;)
pmichaud #bs
oh, wait.
:)
jnthn ;-) 19:54
spinclad #ORANGEbikeshed
pmichaud #bikeshed-champagne
ah, heck, let's just call it #phasers
moritz_ lol
Tene #metarakudo
TimToady you're fired
moritz_ +1 to #phasers
it's free
pmichaud
.oO( TimToady is now TimTrump ?)
19:55
TimToady if yer a phaser, I'm s'posed to fire you
pmichaud +1 to #phasers here, also 19:56
this is where we meet to move Rakudo from one phase to the next :)
PerlJam #phasers wfm (I'm rarely around at the right time anyway :)
TimToady set phasers to #stun, or #deep-fat-fry
colomon +1 to #phasers, though a bit worried we might attract the wrong sort with that name. ;)
spinclad all right, +1 to #phasers here, too
jnthn wfm :-)
TimToady what, people coming back through wormholes?
jnthn Though in a few years somebody will be like...wtf is it called this? 19:57
:-)
[particle] can irc channel names have dashes?
#rakuda-do
moritz_ they can
moritz_ just registered #phasers
colomon moritz_++
moritz_ now I just need to figure out how to add privs for others
[particle] they probably have to join to get privs... maybe not 19:58
pmichaud if #phasers doesn't work out, we can always change again :) 19:59
by then, everyone may have left #rakudosketch :)
Tene /msg chanserv ACCESS #phasers add tene OP
etc.
jnthn whoz op?
Tene my policy is to always just add everyone. 20:00
moritz_ 22:00 -ChanServ(ChanServ@services.)- Invalid template name given, use /msg ChanServ TEMPLATE #phasers for a list
21:59 <moritz_> TEMPLATE #phasers
21:59 -ChanServ(ChanServ@services.)- No templates set on channel #phasers.
Tene huh. 20:01
spinclad LessThanHelpful
Tene the chanserv docs lie, I guess.
spinclad is TEMPLATE a metaword there?
Tene oh, um, it works for me. 20:02
but I get that error if I give the args in the wrong order.
moritz_ tried ACCESS #phasers ADD pmichaud OP 20:06
oh
but +R instead of OP worked
Tene I'll /msg you the templates set on another channel 20:07
moritz_ +1
Tene /msg chanserv flags #phasers tene +votriA 20:09
jnthn So, #phasers next time.
Tene there you go
jnthn :-)
jnthn joins it
spinclad eoq :)
Tene maybe +sRf on some people 20:10
colomon moritz_++ for getting tests added while we're still in #rs
jnthn Aye
Are we done with #rs for the day?
moritz_ thinks so 20:11
TimToady .eeyore
pmichaud I'm done for today.
colomon wfm
pmichaud afk, walk
Tene so everyone clear out of #rs?
20:11 Tene left
jnthn OK, the end 20:11
:-)
spinclad (we can shake out ops grants over in #phasers) 20:12
20:12 TimToady left
jnthn tosses his gavel out, and will use a phaser next time 20:12
20:12 [Coke] left, colomon left, jnthn left 20:13 moritz_ left