11:55 colomon joined 18:43 masak joined 18:59 bkeeler joined
moritz_ oh hai 18:59
jnthn Rakudo Sketch Time!
bkeeler hallo there 19:00
colomon o/
jnthn Wow, quite a few of us.
masak?
19:00 [Coke] joined
jnthn OK, so, we may as well get started. 19:01
Since last #rs, we had Rakudo release thanks to moritz++. I've started hacking on the lexical classes and roles task in the roadmap.
Basics work, non-basics are explodey. 19:02
Fixed various other bugs. Got sorear++'s patch in so we can do Perl 5 module uses, and eval Perl 5 code using Blizkost.
The closeness grows. :-)
moritz_ sounds like a start \\o/
colomon we got a lot more tests passing in the last week. 19:03
moritz_ tried to get s:g/// parsing - failed
jnthn Yes, I was a little lost on where they all came from. :-)
colomon: How's numeric work going? Seems like some good progress?
moritz_ worked on Match objects - blocking on having rakudo building on latest parrot 19:04
jnthn moritz_: Ah, that's a blocker. That means I probably should get on it.
colomon jnthn: I think you unblocked my Role bugs. Or at least, everything I've tried since your patches has worked.
jnthn Wow. I wish everything I tried since my patches worked.
colomon moving slower than I'd like on Numeric, but hoping to have a Rakudo day later this week.
jnthn Great. Did you get unblocked on the weird cmp issue earlier? 19:05
colomon jnthn: not at all, I just reverted it.
jnthn I didn't get chance to backlog and see how that concluded, if it did.
OK.
colomon at the moment it just means there are two tests we don't pass.
jnthn *nod*
colomon but it is full of weird and needs looking into at some point. 19:06
jnthn I'd be interested to get to the bottom of it at some point.
masak is here now
jnthn Especially as it might point to a multi-dispatch bug, though the dump of the candidate sorting didn't look wrong.
yayitsmasak!
masak sorry I'm late. had to pop in at the store before it closed.
jnthn moritz_, colomon: any other bits to report?
colomon I did a bunch of little stuff, like defining more zero-arg versions of operators and getting minmax up and completely redefined.
jnthn colomon+++ 19:07
moritz_ jnthn: updated release guide...
colomon Also +$a now calls $a.Numeric unless there are parrot-y reasons it shouldn't.
jnthn Oh, that sounds like a good step. :-)
moritz_ now contains instructions to list all contributors since the last release
jnthn moritz_++
moritz_ added a perl5 script which helps extracting those names out of the git log
jnthn OK, that'll be a nice touch.
[particle] congrats on the good gsoc news, all. 19:08
jnthn Yes, indeed.
moritz_ (inspired by the perl 5 release announcements, btw)
masak++
pmurias++
and jnthn++ and ruoso++ for mentoring
colomon oh, the other thing I've been at is doing p6 programming for $work. I've run into a couple of minor bugs in the process, but on the whole Rakudo has been impressively stable.
moritz_ EOR from me
jnthn I was especially happy to see that all of the Parrot ones also look interesting for Rakudo.
masak I haven't done much of note in the past week. just submitted 14 rakudobugs. 19:09
colomon masak++ pmurias++ indeed!
masak yay gsoc!
jnthn colomon: That's good to hear.
masak: Any more cheese news?
masak: How much are you tending to use master these days, and how much alpha? 19:10
masak jnthn: GGE is pretty close to working on master. with a bunch of modifications, that is.
mberends++ got proto working on master.
jnthn Great. 19:11
masak: What are the main pain-points on master, from your perspective?
masak November, Druid, Web.pm, Yapsi, Tardis still on alpha.
jnthn (note, answers welcome on this question from others too)
masak jnthn: the main workaroundable pain-point is named enums.
moritz_ JSON::Tiny blocks on Match objects (nothing new here)
masak for projects that depend on named enums, it's quite a large workaround. 19:12
jnthn masak: How much of named enums do they need?
masak from an app-cheesemaker perspective, it feels like a large step backwards, larger than usual.
jnthn: not much, usually.
just the names, in most cases.
jnthn The trouble is that...the spec reached a point of sanity, and then now seems to have complexified again. :-( 19:13
masak there's only one piece of code that I know of that uses the other stuff, and that's the poker script. I already workarounded that one.
jnthn With the dreaded has enum $.blah <foo bar oh ffs why>;
masak jnthn: I never understood why that changed anything.
jnthn Well
The spec doesn't say how it works.
So I can't know even know if it changes anything. 19:14
masak let's postulate it's equal to 'my enum ...' plus an accessor method.
jnthn An accessor method that returns...?
masak that doesn't sound so complex to me.
something sensible.
jnthn Well no, but it doesn't seem so consistent with has-semantics either. But if that's all it is, well, fine.
masak the rationale was to make the enum accessible to deriving classes. 19:15
that's done with the accessor.
jnthn Ok. My feeling is that basically if I implement named enums enough to cover the most basic bits, it unblocks most things.
masak that's right.
jnthn I may do a first cut in cheats
That does enough. 19:16
masak \\o/
colomon \\o/
jnthn OK, so seems I have to fix Rakudo on latest Parrot and do something on named enums. :-)
I'll make those two my priorities 19:17
colomon We still need to do List/Array obviously, but the last week has seen a ton of issues crop up with postcircumfix:<[ ]>
jnthn And then after them return to the lexical classes and roles.
colomon: I noticed.
colomon There are postcircumifx:<[ ]> operators scattered everywhere, each with their own set of features.
moritz_ didn't pmichaud want to attack arrays/lists?
masak should we be concerned about the current state of List/Iterator implementation?
colomon I haven't been able to make head or tails of them.
jnthn colomon: I was hoping pmichaud++ might just be able to make it tonight.
I know he planned to spend the weekend thinking over those lot. 19:18
I'm not sure how far he got.
colomon masak: well, they need to be fixed, but pmichaud should be on them any day.
masak sounds good.
jnthn masak: Perhaps, but since Pm has them as his first priority ATM, I'm not going to panic just yet.
If we get to the May releae and it's not sorted, I'll panic lots though.
colomon I just wanted to shine a light on the related issue, which seems like it will also need work / organization. 19:19
jnthn colomon: On postcircumfix:<[ ]>, I did discuss with Pm this bug a bit.
colomon (And which could probably be done by someone other than pmichaud once he has the big fix in.
)
masak pmichaud is the architect on that front. if he finds he doesn't have time to focus on it himself, maybe the second best thing'll be to leave a design document of some sort, so that the rest of us can try to implement it.
jnthn masak: Yes, agree.
colomon: In theory I could do some changes related to it in advance of Pm's other changes 19:20
But I don't know what he has under way.
And don't want to create conflicts
Since he'll almost certainly be touching those bits of code too.
colomon And I don't know that it's blocking anyone, it just seems to be something that everyone hits when they try messing around with Positionals. 19:21
jnthn *nod* 19:22
Yeah, I understand why it goes wrong now.
It's part of the reason that I decided to do hash slices differently to the way array slices are currently implemented.
It's just a minorly tricky refactor to get array slices that way and touches quite a few places. 19:23
Let's be patient for some more days and see what pmichaud++ has up his sleeve.
colomon agreed.
jnthn If we're in the same position in a week, I'll have a look at fixing it myself.
masak I've heard git branches are a good way to mitigate conflicts, by the way. :) 19:24
jnthn masak: The trouble is more semantic conflicts. 19:25
masak: Like, if I take the code in one direction, and it turns out Pm has taken/is taking it another way.
masak ah, ok.
jnthn I'm not worried about merging at the source control level, more at that level. 19:26
On "lexical variables in regexes" - I think Pm and bkeeler++ had some discussion during the last week.
bkeeler: About?
moritz_ has to leave now, sorry
bkeeler Yep. I have a way forward
jnthn moritz_: No worries, and was good to have you. :-) 19:27
bkeeler Problem is I've had something of a tuitectomy lately
jnthn That sounds painful!
bkeeler $work, spring cleaning, party prep for my 40th birthday
jnthn :-)
Have a fun party. :-)
bkeeler I have next week off though, so I should be able to get to it soon I hope
May 8th if anyone wants to come! 19:28
jnthn I might still be stranded in Iceland. ;-)
moritz_ US?
bkeeler Aye, Portland, OR
moritz_ a bit far :(
jnthn That's...a long way. :(
But thanks. :-) 19:29
bkeeler Hehe, maybe next year
moritz_ gone&
jnthn Anyways, glad that things are unblocked and there's a way forward. :-)
bkeeler Indeed
ball's firmly in my court
jnthn :-)
On the autovivification, I have a better idea of what needs doing now. It sounds mildly tricky and has some easy-if-you-love-the-binder-and-guts dependencies, probably. 19:30
masak :) 19:31
jnthn But I feel less blocked on it than I did a week ago.
"improved error messages and failure modes" - who wants to write the backtrace printer? ;-) 19:32
Apparently there's Parrot improvements on line number reporting though, so I'm hopeful that this will mean once those go in, the line numbers will be decent.
bkeeler So what's the story on the parrot blockage? 19:33
Something to do with :nsentry?
jnthn Yeah
bkeeler What's that do? Anywhere I can read up on it?
jnthn We were depending on something that (we asked to be and) got deprecated.
Well
s/were/are/
:nsentry lets you specify the name something in installed in the namespace under. 19:34
Or that it gets a namespace entry.
bkeeler Ah
jnthn I think the code that needs fixing may (though may not) be moderately hairy. :-) 19:36
It needs a proper fix though rather than scattering :nsentry over the whole codebase.
colomon: What's left, in your view, before we can mark: 19:37
2 ** true hyper/cross/reverse/other metaoperators
As done?
colomon hmmmm
jnthn What we have now feels...pretty true. :-)
masak ship it!
colomon there are still some fudges
but it's a lot closer than it was, that's for sure. 19:38
jnthn fudges as in, in the tests?
masak colomon++ moritz_++ jnthn++
jnthn Or in the implementation?
colomon in the implementation.
the tests are still woefully lacking, I fear.
jnthn OK.
I'll leave it in the roadmap as an in progress task then. 19:39
Just caught my eye as a "are we there yet" :-)
colomon at the same time, I don't know that I'd be too upset if they went out as is.
hyper on hashes probably should get done before R* 19:40
jnthn Ah, yes.
It used to work in alpha.
colomon we don't have the Sop at all.
jnthn Yeah but
colomon mostly because it mystifies me.
jnthn We don't have parallelism at all.
So Sop is, afaict, pointless
colomon by that logic, Sop should be a nop, and easy to do.
:)
jnthn I don't consider that a blocker to us marking the task complete.
lol 19:41
:-)
Well
colomon but S has different effects on things like &&, doesn't it?
jnthn Yes, but that'll make our life harder when we come to do parallelism and actually have to give it meaning since it's already there. ;-)
Possibly
I say we just skip it for now.
colomon I won't argue with skipping it.
masak +1
jnthn Who knows, if we ignore it and fail to comprehend it enough, it may just go away. :-) 19:42
OK, any more points for this week's #rs? What are folks hoping to hack on this week? 19:43
(and are there any blockers for it that we didn't already discuss) 19:44
[particle] what's the latest on TT#389? 19:46
colomon I'm hoping to dive into the heart of the Numeric stuff. But first I need to write a bunch of tests for log and exp first, to work on sorting out that everything works right if you, say, call Complex.log with a Rat argument.
[particle] point me to the log if i missed it
jnthn [particle]: Well, the ticket is closed by the looks of it. :-)
[particle]: But Rakudo hasn't caught up with the change yet. 19:47
I may be able to fix that tonight.
[particle] ah, there it is.
jnthn If not, probably tomorrow.
colomon: OK, sounds good.
[particle] if you do, we can merge compact_strings_revamp, and you'll get ~10% more speed
jnthn Should the immutable strings have got us a win too? 19:48
yay 10%!
colomon Of course, I'd also like to see us break 32,000 passing tests, and if a little random Rakudo hacking is needed, so be it. :) 19:49
jnthn ;-)
OK, if that's about all we have, I think that'll do for this week's #rs. 19:51
Thanks everyone!
bkeeler \\o
colomon o/
19:51 bkeeler left 20:18 masak left

Powered by Cro and the Raku® Programming Language.

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

Thank you!