Perl 6 - the future is here, just unevenly distributed

IRC log for #rakudosketch, 2010-05-11

| Channels | #rakudosketch index | Today | | Search | Google Search | Plain-Text | summary

All times shown according to UTC.

Time Nick Message
03:07 TimToady joined #rakudosketch
04:58 sorear joined #rakudosketch
10:14 masak joined #rakudosketch
18:30 diakopter joined #rakudosketch
18:55 diakopter sketchy
18:55 moritz_ aye
18:56 masak o/
18:56 masak still a few minutes left.
18:56 colomon t minus 4
19:00 jnthn o/
19:00 moritz_ ¸o
19:00 jnthn .oO( role Call; )
19:00 masak :P
19:00 moritz_ that's me hanging over the chair because I ate too much for dinner
19:00 jnthn Sorry I wasn't about last week - the wilderness has no internets.
19:01 spinclad \o
19:01 moritz_ np
19:01 colomon o/
19:01 jnthn I've nothing much to report - I comitted a couple of minor fixes, but mostly spent my vacation being, well, on vacation. :-)
19:02 colomon I did a bunch of Numeric stuff: sqrt, log, sin, comparison operators (mostly for Real) so far.
19:02 jnthn OK, sounds good.
19:02 moritz_ anythin' you're blockin' on?
19:02 colomon Plus Sop, cmp for Pairs, and fixing ^$n so it makes $n be a number.
19:02 colomon I don't think so.  In particular, getting sin done means that all the trig functions are just a matter of time.
19:03 moritz_ that's good news
19:03 jnthn colomon++
19:03 jnthn Sounds like good progress both for Rakudo and your grant.
19:03 moritz_ indeed colomon++
19:03 colomon I'
19:03 colomon I'm assuming I can't do TrigBase $base = CALLER::<$?TRIGBASE> yet?
19:03 moritz_ nope
19:03 jnthn No
19:03 moritz_ pseudo packages NYI
19:03 jnthn Will you block on that?
19:03 colomon I don' think that's really a blocker.
19:03 jnthn OK.
19:03 colomon just a detail that can't be implemented yet.
19:04 jnthn pseudo packages are on the ROADMAP
19:04 moritz_ I applied some patches
19:04 moritz_ lured another contributor into patching tests
19:04 masak moritz_++
19:04 moritz_ not much progress with match objects...
19:05 moritz_ the problem is that NQP-rx uses vtable calls to populate the Match object
19:05 jnthn Yay, another test hacker. \o/
19:05 masak I don't have much to report either. dry spell with rakudobugs right now. will probably get started on my GSoC grant as soon as I'm done with Moscow+presentation.
19:05 masak a week from now.
19:05 moritz_ and for some of them I've only encountered recursions
19:05 colomon \o/
19:06 moritz_ between set_pmc_keyed and postcircumfix:<{ }> in Associative
19:06 jnthn moritz_: Yeah, I can see how that could end up happening.
19:06 moritz_ jnthn: any good ideas what I can do about it?
19:06 jnthn moritz_: Will have to look at the details to see a solution.
19:06 moritz_ ok
19:06 moritz_ I'll brush up the branch and push it
19:06 moritz_ I have a hackish workaround
19:07 moritz_ that calls eval_set_pmc_keyed method
19:07 jnthn moritz_: OK, sounds great. I'm not too optimistic about getting time today, but tomorrow after I've given my talk should be better.
19:07 moritz_ which I've then implemented in Match
19:07 moritz_ I've got non-quantified named captures working that way
19:07 moritz_ more of a proof-of-concept, nothing to go into rakudo
19:07 jnthn *nod*
19:07 jnthn masak: Off to Moscow this week?
19:08 moritz_ that's my report, basically
19:08 masak jnthn: yes. Sunday, ashcloud permitting.
19:08 jnthn masak: Yay! And I hope the ashcloud permits.
19:08 masak me too.
19:09 jnthn masak: Anything you're blocking on or needing in the GSoC run-up?
19:09 masak to complicate matters, I'll be in Värmland the two days prior to that.
19:09 masak jnthn: don't think so. I could round up a bunch of 'serious' RT tickets, if that helps.
19:09 masak but I'm not blocking on the grant yet.
19:10 jnthn masak: Having some user perspective on what tickets hurt most is a nice thing.
19:10 moritz_ anybody else got something to report?
19:11 moritz_ if not, i'd like to turn our attention the to the book
19:11 moritz_ I've added somethiing about comparison operators, and plan to add more about string comparison ops and smart matching
19:11 jnthn OK, sounds good.
19:11 moritz_ szabgab expressed interest in contributing... let's see how much time he can invest
19:12 jnthn I plan to do more book than code in the next couple of weeks.
19:12 moritz_ do we have any priorities for getting it printable?
19:12 colomon I probably could have used whatever you wrote, I totally botched the lt, le, gt, etc operators over the weekend.  :)
19:12 moritz_ :-)
19:13 moritz_ I've set the book title to 'Using Perl 6', and will keep it that way until sombody protests
19:13 jnthn moritz_: The OO and roles chapters are still weak. subs-n-sigs needs a bunch more work.
19:13 jnthn moritz_: Those are the areas I expect to mostly focus on.
19:13 moritz_ ok
19:13 moritz_ also dropping in TODO markers where you feel somethiing is weak is already helpful
19:14 jnthn moritz_: Yes, certainly.
19:14 jnthn moritz_: It'd be good to grep through for the ones we already have.
19:14 jnthn (That's for all of us, not just moritz_.)
19:14 moritz_ did that last weekend, and eliminated a few of them
19:14 masak I'll focus more on the book too, starting next week. I still think we need more chapters, but I'm not sure what more needs explaining.
19:14 jnthn To get a feel for other bits that need improving.
19:14 jnthn masak: I felt more that we need to fill out many of the chapters we already have some more.
19:15 moritz_ I'd also like to get some more feedback from chromatic - it has been hard to lay hand on him during the past few days
19:15 masak I still think a poker hand example would be good for explaining subsets.
19:15 masak jnthn: oh, definitely.
19:15 jnthn masak: Do you have time to write the code example?
19:15 masak jnthn: I think so.
19:15 * jnthn finds it much easier to write about and explain examples than to come up with them, fwiw.
19:15 moritz_ that would be great; others could then chime in and... right
19:16 masak jnthn: there's already working code.
19:16 masak I could put it in the book repo and see what happens. :)
19:16 jnthn Go for it. :-)
19:16 * masak does so
19:16 moritz_ bad things, hopefully not, speaks yoda
19:16 jnthn A lot to write still we have.
19:17 masak why suddenly like this we speak?
19:17 moritz_ of the yoda conditions /me thought
19:17 jnthn To get it out of us now so it doesn't end up in the book.
19:18 masak :)
19:19 jnthn I probably need to take the time at the weekend to read through the book that we have so far to get a feel for what we have / haven't got.
19:19 jnthn As well as writing some bits.
19:19 * jnthn doesn't have a good feel for the overall state ATM.
19:19 moritz_ if you have some gripes about rendering in latex, talk to me
19:19 [Coke] (poker hand example) ah, I think I still owe someone code on that.
19:20 moritz_ I've hacked up the latex emitter quite a bit
19:20 masak [Coke]: would be nice if you could look at the XXX comment. http://gist.github.com/244255
19:20 moritz_ the only remaining issues I know are side bars
19:20 masak [Coke]: if you don't have time, that's cool too.
19:23 jnthn Any more bits on book?
19:23 moritz_ don't think so
19:23 jnthn OK.
19:23 jnthn Any more things to discuss on Rakudo * bits?
19:23 jnthn I notice sorear++ sent in a patch for one of our ROADMAP priority 1 items.
19:24 jnthn I didn't read pmichaud's feedback yet; I did see things in it that made me uncomfortable
19:24 jnthn But the fact that we have a patch that works is a good starting point.
19:24 moritz_ aye; pmichaud commented that the ground work was solid, but more polishing required
19:25 jnthn Ah, OK, that's promising then.
19:25 * moritz_ has high hopes
19:25 moritz_ I plan to assemble a list of most annoying bugs that are my personal favorites for 'fix before R*'
19:26 masak are we 'on track' for the priority 1 items?
19:26 * moritz_ is a bit worried about lists/arrays
19:26 jnthn Ah, also (certainly since the last Rakudo sketch I was at) we also got "lexical classes and roles" marked as complete.
19:27 jnthn moritz_: Me too - I see in the Perl 6 call pmichaud still expects to have time for those soonish.
19:27 jnthn "improved error messages and failure modes" - I'm quite happy we've made big steps forward on our parse time errors here.
19:28 jnthn I may try and get Perl 6 level backtrace printing done again.
19:28 jnthn The problem with that item is that we can always be More Awesome
19:28 masak that's not a problem :)
19:28 masak except about us being unable to mark it as finished, of course.
19:29 jnthn But I think if we can get mostly-correct runtime errors nailed, then I'd be prepared to say that we're done enough.
19:29 jnthn That is, ones that report the line/file in the Perl 6 source.
19:29 jnthn "lexical variables in regexes" - don't think bkeeler is about ATM to ask; anyone know anything on that?
19:30 masak he made patches quite a while ago for those.
19:30 masak don't know what heppened since, only that they didn't go in.
19:30 jnthn OK, since pmichaud++'s last round of feedback?
19:31 masak no idea.
19:32 jnthn OK
19:32 * colomon is also worried about closures.
19:32 * spinclad worries about closures -- they're getting pm's attention, working with #parrot -- /me wonders if i could talk with pm
19:32 moritz_ spinclad: did you see the discussion on #perl6 today about closures?
19:32 jnthn Yeah, those are the two bits that we're currently quite dependent on pmichaud++ for
19:33 spinclad moritz_: the start of it, at least, still backlogging
19:33 colomon three bits?  lists, closures, and variables in regexes?
19:33 moritz_ sounded like he had a plan forward (on closures), at least partially
19:34 jnthn colomon: I'm not sure if variables in regexes are currently blocking on another round of feedback or not.
19:34 jnthn moritz_: Yes, I believe so.
19:34 moritz_ I think on bkeeler's tuits
19:34 jnthn OK.
19:35 moritz_ (which he announced three weeks ago, or so)
19:37 jnthn OK. So that's most of the priority 1 ROADMAP items discussed. I'd really like it if one more got knocked off before #rs next week.
19:37 jnthn Or more than one.
19:38 jnthn I can probably take at least one of them. :-)
19:38 jnthn Do we have any progress on getting the advent calendar examples into some kind of a test sutie?
19:38 jnthn *suite
19:39 jnthn 1 ***   get the Advent examples running again (all)
19:39 jnthn I'd really like to know how close we are/aren't.
19:39 colomon has anyone done anything at all for that? (he asked, embarrassed....)
19:40 jnthn colomon: Not that I'm aware of.
19:41 moritz_ I've gone through some of them
19:41 moritz_ the first posts don't need any rakudo updates
19:41 jnthn moritz_++
19:41 moritz_ but I forgot how far I got
19:42 jnthn I figure the ones towards the end are likely to get harder.
19:42 moritz_ all issues I've seen were known/ticketed
19:42 jnthn Since we covered more advanced stuff later on.
19:46 jnthn OK, anything else anyone wants to discuss?
19:46 jnthn Otherwise, we'll wrap up here and I'll go cook my noms.
19:46 moritz_ good nomming
19:47 colomon I kind of feel like we could use a Rakudo pep talk. (which I definitely don't have in me.)
19:49 masak for what it's worth, I still think we can put forth something quite good for June.
19:50 jnthn Me too. Rakudo certainly has its weak areas, but it also has a lot of strong ones too.
19:50 colomon woah.  I don't think I'd actually conceptualized it before, but I think that's it -- I'm starting to get worried about June.
19:50 masak it's mostly a question of packaging and tutorials and examples at this point.
19:51 * moritz_ sees lots of stuff that should get fixed first
19:51 colomon I'm with moritz_.
19:51 jnthn I think that for June, we're certainly not on course to deliver something perfect - but that was never The Plan. If we try to do that we'll never release anything. We're on course to deliver an implementation that covers many of the things that Perl 6 has promised.
19:52 jnthn Yes, we've got some nasty bugs and rough edges still.
19:52 jnthn But we've also fixed a lot of nasty bugs to date.
19:52 moritz_ aye
19:53 jnthn A lot of things are coming together. There's certainly some very concerning areas, but there's a lot of things that we're doing really quite well.
19:53 colomon I think what has me worried is it appears that many of the remaining nasty bugs are blocking on pmichaud's tuits.
19:53 jnthn colomon: Yes, that is certainly a concern.
19:53 moritz_ it would be nice to find a way to increase our bus number
19:53 jnthn It's my biggest worry too.
19:53 * diakopter is curious why sorear hasn't spoken up
19:54 jnthn I've already picked off many of the eaiser things that would have blocked on pmichaud (like the user defind operators and item assignment)
19:54 spinclad what's the limit on handing the book off?  that's the tightest constraint now, maybe
19:54 jnthn But the lists bit is, well, rather harder to Just Do.
19:54 jnthn spinclad: AFAIK, end of the month.
19:55 diakopter imho, sorear himself increased the bus number significantly.
19:55 colomon end of May?
19:55 jnthn diakopter: I think timezone may be part of it.
19:55 diakopter he's in PDT like me I think
19:55 jnthn colomon: Aye, thus my mention earlier that I intend to spend more time on book than code in the next while.
19:55 jnthn diakopter: Physical timezone and actual timezone don't always correlate. ;-)
19:56 colomon sorear++ has been doing fantastic things. but they don't seem to be making much of an impact on Rakudo yet.
19:56 jnthn But yes, sorear++ is very much digging into Rakudo.
19:57 jnthn colomon: They are, and will. I expect once the pollishing is done, the patch for the REPL bits will go in, for example.
19:57 moritz_ colomon: my first 20 patches to rakudo didn't make much of a difference to rakudo, not each on its own
19:57 jnthn colomon: And the patches for the first bits of Perl 5 support did also.
19:58 colomon sorry, that wasn't meant to be a knock on his patches.
19:58 jnthn colomon: I didn't read it was one, fwiw.
19:58 colomon more a remark on how they are stacking up in a corner by themselves instead of becoming part of master.
19:59 colomon I don't think that's sorear's fault, either.
19:59 moritz_ it's more lack of patch applier's tuits
19:59 jnthn Well, patch reviewer's tuits also.
19:59 diakopter my thought on that ("more a remark..") is that sorear should be encouraged to blaze trails notwithstanding patch reviewers' availability
19:59 moritz_ +1
20:00 spinclad +1
20:00 diakopter (sorry if I'm stating the obvious) :|
20:00 * diakopter shuts up
20:00 colomon +1
20:00 spinclad (obvious)++
20:00 colomon I just want the rest of us to be able to follow those trails.
20:01 masak +1
20:01 jnthn +1, and I'll try to do my part to be reviewing patches from him too
20:01 masak I really like how sorear++ has been blazing trails so far.
20:01 masak he's really upsetting the status quo :)
20:02 jnthn Yes. The macros work was a big surprise. :-)
20:07 pmichaud hello
20:07 jnthn \o/
20:07 jnthn pmichaud: hi!
20:07 pmichaud sorry I'm late -- small disaster here.
20:08 jnthn pmichaud: :-(
20:08 spinclad \o/
20:08 masak welcome, pmichaud.
20:08 jnthn pmichaud: No worries, happy you've made it at all. :-)
20:08 colomon o/
20:08 spinclad (welcome, glad you made it)
20:08 pmichaud (long story short... while installing a new cable so that we can get better cell coverage in the house, I hit a water line)
20:09 pmichaud OOOPS
20:09 jnthn D'oh!
20:09 * masak sucks in breath
20:09 colomon yow!
20:09 pmichaud although why the builders stuck a water line mere inches away from an electrical outlet is beyond me.
20:09 * spinclad hold breath, treads water
20:09 pmichaud anyway, closures.  I should have closures fixed in the next day.
20:10 jnthn .oO( Quits pmichaud: excess flood )
20:10 spinclad \o/
20:10 colomon \o/
20:10 jnthn pmichaud++ # yay!
20:10 colomon that's huge.
20:10 * moritz_ wondered for a second if "closures" can be used to close water leaks
20:10 pmichaud Fortunately I took a shortcut when working on Parrot lexicals back in 2008 that means that fixing closures will be far easier than if I had done it the right way.
20:10 colomon shortcut++
20:11 pmichaud essentially, capture_lex and newclosure don't do as much checking as they probably should.  They're a bit naive.  And the fact that they're naive means I can fix them.  :-)
20:11 pmichaud s/fix them/exploit their naivety to make sure that closures actually work/
20:11 pmichaud although we may have trouble with closures that are also multis
20:11 jnthn We'll just call it an "extensible implementation"
20:12 pmichaud but I don't think we have many closures that are multis.  And it might work out correctly anyway.
20:12 pmichaud anyway, I'll have closures fixed in the next day or two.
20:13 moritz_ we don't?
20:13 jnthn pmichaud: iirc, last time you hit some issues with parametric roles when doing closure fixes. If they come down to your only blocker, but everything else works and the other bugs are fixed, feel free to punt the fixes to those to me.
20:13 pmichaud jnthn: I think it's all going to Just Work.
20:13 masak \o/ # closures
20:13 pmichaud jnthn: but yes, if I run into problems I'll punt.
20:13 colomon do we have good tests for the closure-related bugs?
20:13 jnthn pmichaud: OK. And if it Just Works that'll be wonderful. :-)
20:14 pmichaud I think it'll just work because PCT is smart enough to recognize when we're using blocks as rvalues :-)
20:14 spinclad (vs ivalues (invoke/immediate))
20:18 pmichaud well, it's not only then
20:18 pmichaud if we have
20:18 pmichaud {   say 'hello';  sub foo() { ... };   say 'xyz'; }
20:18 pmichaud then foo() is not immediate but it's also not being used as an rvalue
20:19 pmichaud (i.e., it's in sink context)
20:20 pmichaud although it might be a bit tricky because the block is bound to lexical &foo.  Anyway, it'll either just work or we'll have to muck around with lexicals just a bit.
20:21 pmichaud I'm hoping for "just work" (although perhaps sub-optimally for a short bit)
20:21 pmichaud reading backlog
20:21 moritz_ well, anything better than what we have right now is an improvement
20:21 * spinclad wants to know more but doesn't want to devote #rs to it.  could i ask you about back on #p6 after?
20:21 pmichaud I call "tautology!"  :-)
20:21 moritz_ pmichaud: I was quite aware of that :-)
20:22 diakopter I call truism!
20:24 pmichaud anyway, my #rs report:  I've been working on fixing closures, should have that in the next couple of days.
20:25 pmichaud things are settling down around the house a bit so that as expected, I should have more time to devote to R*
20:25 pmichaud I'm currently catching up the spectest-progress.csv file on my notebook
20:26 pmichaud after closures are fixed, I'll work on lists/arrays next
20:26 pmichaud I'll also see if I can make some better pointers about how interactive mode should work -- the current patch is too brutal on the code generators
20:27 pmichaud EOR
20:33 * jnthn will be very relieved to see closures fixed. :-)
20:33 * spinclad too
20:34 masak oh yes.
20:53 spinclad meeting over?
20:54 masak seemingly.
20:54 spinclad <gavel>
21:02 masak left #rakudosketch
21:13 diakopter left #rakudosketch
23:25 spinclad phenny: tell ash__ (even if thrown away in sink context) ... but pmichaud seemed to have a different point of view on #rs.
23:26 spinclad EWORNGWINDOW

| Channels | #rakudosketch index | Today | | Search | Google Search | Plain-Text | summary