00:15 vendethiel joined 00:55 vendethiel joined 01:34 Peter_R joined 01:36 vendethiel joined 02:55 vendethiel joined 03:19 ShimmerFairy joined 03:48 vendethiel joined 04:12 vendethiel joined 04:43 vendethiel joined 05:19 vendethiel joined 06:16 vendethiel joined 06:22 ggoebel joined 06:48 FROGGS joined 06:49 Ven joined
dalek arVM: 7dfa89b | skids++ | src/strings/ (3 files):
Update windows-1252 codepage support and activate it.
 06:57
arVM: 060346c | skids++ | src/strings/windows1252.h:
Add missing C prototype.
arVM: bb50941 | FROGGS++ | src/strings/ (4 files):
Merge pull request #210 from skids/windows-1252

Windows 1252 support
07:13 vendethiel joined
jnthn FROGGS: Commented; thanks :) 07:34
FROGGS jnthn++
:o)
jnthn makes coffee and prepares to dig into Perl 6 things :) 07:42
FROGGS \o/
sounds good :o)
btw, my attempts to nuke an SC were not successful yet, though, I won't give up that quickly 07:43
jnthn Hmm 07:44
07:46 vendethiel joined 07:48 zakharyas joined 08:07 rurban joined
dalek arVM: 826a8cc | jnthn++ | src/strings/unicode_ops.c:
Fix uniname out of range error reporting.
 08:15
jnthn eek, turns out RT #124333 is actually a bug in computing SSA form. 08:24
synbot6 Link: rt.perl.org/rt3//Public/Bug/Displa...?id=124333
FROGGS ohh deer 08:26
dalek arVM: ae63692 | jnthn++ | src/ (3 files):
Fix SSA crash when huge number of basic blocks.

When we have a lot of basic blocks, we might end up needing to produce huge PHI nodes. This was exposed in real world code of only 8KB when a huge alternation was written in a regex. This fix both bumps the size of the field holding operand count so we can handle such large graphs, and adds a check so we panic rather than SEGV in the (rather unlikely) event we ever exceed the new limit.		 09:01
09:06 vendethiel joined 10:00 vendethiel joined
dalek arVM: 8bcd46a | jnthn++ | src/strings/normalize.c:
Implement normalization buffer grwoth handling.
 10:10
10:13 japhb joined 10:27 vendethiel joined 11:05 vendethiel joined 11:28 vendethiel joined 11:37 Ven joined
dalek arVM: d8d853b | jnthn++ | src/strings/utf8.c:
Add missing frees on invalid UTF-8 decoding.
 11:45
11:53 vendethiel joined
dalek arVM: cc08ed5 | jnthn++ | src/strings/utf16.c:
Integrate NFG with UTF-16 non-stream decode.
 11:56
11:58 lizmat joined
dalek arVM: 0bff027 | hoelzro++ | / (13 files):
Remove linenoise/readline

Now that NativeCall is included with Rakudo, instead of having linenoise/readline functionality be bundled in as a VM-level op, let's create a NativeCall-based module for this functionality		 12:13
arVM: 343ff5b | hoelzro++ | / (13 files):
Merge branch 'no-moar-linenoise'
13:03 vendethiel joined
dalek arVM: eff0e68 | jnthn++ | src/strings/decode_stream. (2 files):
Give a decode stream a normalizer.

Up to the encodings if they will actually use it.		 13:07
arVM: dc9a9cd | jnthn++ | src/strings/decode_stream.c:
Further integrate normalizer with deocde stream.

We're not finished if there are still graphemes to retrieve from the normalizer, and we need to EOF the normalizer when told to fetch the remaining graphemes.
arVM: 660cef5 | jnthn++ | src/core/interp.c:
Toss unrequired curly clutter.
 13:38
arVM: 6139281 | jnthn++ | src/strings/ops. (2 files):
Fix chr for our brave new NFG world.
13:51 vendethiel joined 13:52 Ven joined 14:15 vendethiel joined
[Coke] is there a way to make moarvm.org/releases/ list releases in descending order? 14:25
nwc10 <joke>descending order of what? Awesomeness?</joke> 14:26
which I figure is actually 1 to 1 with date
14:26 Ven joined
jnthn [Coke]: Maybe, it's a normal Apache web server, if there's a line I can drop in a .htaccess that'll make happen then it's easy enough 14:27
dalek arVM: 4108a85 | coke++ | ports/macports/Portfile:
Update macportfile to 2015.04 release
 14:29
arVM: c2b904f | jnthn++ | src/strings/normalize. (2 files):
Make \n a "normalization terminator".

This means we always assume we won't get any combiners after it, and so do not need to block until we get another codepoint. Makes REPLs and similar keep working once we switch NFG on for all strings.		 14:40
14:40 Ven joined 14:42 vendethiel joined 15:15 vendethiel joined
TimToady perhaps all control codes should be normalization terminators? 15:37
thinking of things like tab completion... 15:38
or synchronizing with ESC interpretation 15:40
jnthn TimToady: Possibly. 15:41
TimToady: I coded it expecting this not to be the only one.
But didn't have a good idea just how many we'll end up with. 15:42
TimToady well, < 32 is pretty easy to test for
jnthn Are all the controls contiguous?
Ah. Then yes :)
15:42 FROGGS joined
TimToady m: say uniprop(0x80) 15:44
camelia rakudo-moar 0914fc: OUTPUTĀ«Ccā¤Ā»
TimToady up there too
jnthn ah
The latin-1 ones I ugess
*guess
TimToady m: .say if .uniprop() eq 'Cc' for ^0xffff 15:45
camelia rakudo-moar 0914fc: OUTPUTĀ«0ā¤1ā¤2ā¤3ā¤4ā¤5ā¤6ā¤7ā¤8ā¤9ā¤10ā¤11ā¤12ā¤13ā¤14ā¤15ā¤16ā¤17ā¤18ā¤19ā¤20ā¤21ā¤22ā¤23ā¤24ā¤25ā¤26ā¤27ā¤28ā¤29ā¤30ā¤31ā¤127ā¤128ā¤129ā¤130ā¤131ā¤132ā¤133ā¤134ā¤135ā¤136ā¤137ā¤138ā¤139ā¤140ā¤141ā¤142ā¤143ā¤144ā€¦Ā»
TimToady looks like that's all that's considered Cc 15:46
jnthn OK, then we can do those
timotimo m: say uniname 0x140 15:47
camelia rakudo-moar 0914fc: OUTPUTĀ«LATIN SMALL LETTER L WITH MIDDLE DOTā¤Ā»
timotimo er
m: say uniname 140 15:48
camelia rakudo-moar 0914fc: OUTPUTĀ«PARTIAL LINE BACKWARDā¤Ā»
timotimo those are control characters, eh?
m: say chr 140
camelia rakudo-moar 0914fc: OUTPUTĀ«ā¤Ā»
timotimo i don't think i've heard of those
TimToady oh, and all the ASCII escapes are completely obvious :P 15:49
*controls 15:50
DATA LINK ESCAPE?
timotimo kind of sad to see all those hardly ever used by any programs 15:51
TimToady hey, we could make them into variable names! 15:52
FROGGS \o\ special vars ftw! /o/ 15:54
15:55 Ven joined, vendethiel joined
FROGGS $ perl -e 'my $ = 42' 15:57
Can't use global $^ļæ½ļæ½ļæ½ļæ½ļæ½ļæ½ļæ½ļæ½ļæ½ļæ½ļæ½ in "my" at -e line 1, near "my $ļæ½"
TimToady heh, I can't copy/paste that line because the ^U wipes out everything but the final quote :) 16:00
jnthn hah! 16:01
TimToady (when I produce the line locally, that is)
jnthn TimToady: See #perl6, but we've a few tests that I don't think are gonna work out with NFG.
(7 of thme, but all in the one test file, I *think*) 16:02
(I need to re-run spectest to make sure fixes I'm doing don't bust other things)
dalek arVM: 5e8d2b0 | jnthn++ | src/strings/ (3 files):
Ensure concatenation is closed under NFG.
 16:17
16:28 vendethiel joined 16:46 Ven joined 16:56 vendethiel joined
dalek arVM: 5a47535 | jnthn++ | src/core/interp.c:
Don't leak synthetics from ord.
 17:10
17:23 japhb joined
dalek arVM: 18fcc09 | jnthn++ | src/strings/utf8.c:
Integrate NFG with UTF-8 non-stream decode.
 17:42
arVM: 10f3ad3 | jnthn++ | src/strings/utf8.c:
Implement NFG for decode stream case of UTF-8.

This covers file I/O.
jnthn This basically flips NFG on for those who build master all the time. :)
17:44 vendethiel joined
[Coke] nice. 18:06
trac.macports.org/ticket/47561 18:07
dalek arVM: c799226 | coke++ | ports/macports/README:
more instructions
 18:24
arVM: dcb8d11 | coke++ | ports/macports/README:
Even moar instructions.
[Coke] patch applied, should show up in ports soon 18:26
18:58 brrt joined
brrt \p 19:13
\o
19:34 brrt joined 20:23 colomon joined
brrt hmmm... invalid thread id in gc pass ... that's likely to be a mvmroout problem,no? 20:38
jnthn Quite often
If you actually have no threads (or even if you do) it tends to mean memory corruption.
brrt guess what, it doesn't fail without make
jnthn Then, memory corruption :( 20:39
brrt lets try, try again
hmmm 20:41
ok, i reliably get a broken moarvm now
jnthn OK. 20:42
Well, debugging time ;)
brrt well, i'm checking if it's been brought on by the recent rakudo changes 20:43
jnthn It's almost always a mistake in Moar that's been a ticking time bomb, and sadly expereince tells that knowing a recent Rakudo change brought it on rarely provides any insight into how to fix it. 20:44
brrt well, in this case it could be an bigint in an int 20:48
timotimo do we output the thread id, too? 21:03
because if it's 0, 1, 2, 3, ... it'd be fine, but if it's -12431 or 95028 it'd be pretty problematic
jnthn timotimo: No, but no objections if you want to add it 21:04
timotimo sure, no problem :)
jnthn: does the localref/local thing have a spot in your priority queue yet? 21:05
i mean, now that the release is through, it's not terribly pressing, i suppose
jnthn timotimo: I'll be doing another push on native stuff soon, yes. 21:06
timotimo: Depends how much NFG stuff I have to jump on.
timotimo ... and actually when i tried to build the tests for lexical/lexicalref i couldn't get them to work, but i haven't actually inspected the bactrace i got enough
right
big ups for your nfg work, btw :)
jnthn Our LTM-er is bust in a few interesting ways with regard to NFG, it turns out. 21:07
timotimo oh
it doesn't know anything about nfg yet?
jnthn No, and worse a couple of aspects of the QRegex::NFA design make it a bit hard to fix.
timotimo damn ;( 21:08
jnthn (Not horribly hard, but not "oh yay it just works" either.)
However, we don't yet have a single spectest bitten by this.
Uh, we may have one that I wrote this week. :)
timotimo better to have one than to not have one
jnthn Yeah
Anyway, I'll ponder that some more next week :)
timotimo i ought to write a test for that hyperop thing i found the other day, i think 21:09
jnthn I think it's hard enough to run into that it's very unlikely to cause ecosystem strife.
timotimo should that be filed as todo or left failing?
i haven't roast in a loooong time, it feels like
jnthn The hyperop thing? todo
timotimo good
wat 21:18
there are already tests that look like exactly what is wrong in my ticket 21:19
21:22 rurban joined
[Coke] timotimo: nice, right? 22:01
(or are those tests passing?)
timotimo they do pass 22:02
22:38 flussence joined 22:48 dalek joined 22:59 colomon joined

Powered by Cro and the RakuĀ® Programming Language.

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

Thank you!