Perl 6 language and compiler development | Logs at colabti.org/irclogger/irclogger_logs/perl6-dev | For toolchain/installation stuff see #perl6-toolchain | For MoarVM see #moarvm
Set by AlexDaniel on 12 June 2018.
lizmat Files=1242, Tests=76416, 320 wallclock secs (15.45 usr 5.32 sys + 2218.26 cusr 226.65 csys = 2465.68 CPU) 09:36
[TuxCM] Rakudo version 2018.06-80-gade83c861 - MoarVM version 2018.06-127-g513000322
csv-ip5xs0.925 - 1.013
csv-ip5xs-207.639 - 7.685
csv-parser25.850 - 26.535
csv-test-xs-200.436 - 0.457
test9.129 - 9.374
test-t2.334 - 2.394
test-t --race1.004 - 1.046
test-t-2042.429 - 43.332
test-t-20 --race14.939 - 15.637
10:29
Geth nqp/truffle: 3d22936a62 | (Paweł Murias)++ | tools/build/Makefile-JVM.in
[truffle] Fix Makefile
11:03
Geth ¦ rakudo: zoffixznet self-assigned %(), @(), $() are evil/weird github.com/rakudo/rakudo/issues/1946 12:45
6.d-prep: d87b37c845 | (Zoffix Znet)++ | 8 files
Defer features we can't do in 6.d to 6.e

All the things that R#1289 blocks are now postponed to 6.e
  (or later, if needed). It's a difficult issue that shouldn't
be rushed and we should release 6.d with what we have, which includes 3100+ spec commits.
... (9 more lines)
12:56
synopsebot_ R#1289 [open]: github.com/rakudo/rakudo/issues/1289 [6.e] Implement a Way to Know Caller's Language
6.d-prep: b937374b15 | (Zoffix Znet)++ | deferred-to-6.e/README.md
Commit missing content for d87b37c845cc55632

Had some unsaved content in the editor. This goes with commit github.com/perl6/6.d-prep/commit/d...7c27e6148e
12:58
Geth nqp/truffle: 34c1b32a55 | (Paweł Murias)++ | 3 files
[truffle] Implement nqp::isnull and nqp::ifnull
14:37
Geth nqp/truffle: 199115bc4a | (Paweł Murias)++ | src/vm/jvm/Truffle.nqp
[truffle] Use NYI method for NYI stuff
18:38
nqp/truffle: fe22a03de4 | (Paweł Murias)++ | 6 files
[truffle] Allow @Deserializer to handle arrays of nodes
Geth 6.d-prep: 67789986cb | (Zoffix Znet)++ | ChangeLog.md
Start 6.d ChangeLog

  **Please, don't publicize this file, as people will start
changing their code based on it, but it's incomplete/incorrect and contains stuff that is yet to be done.**
The requirement of at least one implementation (end of text in commit) I just made up right now, but I think it's a good idea and will avoid us promising unicorns in the spec that look good on paper, but end up conflicting with some part of the language when implemented.
18:59
6.d-prep: fbe3c2ffc4 | (Zoffix Znet)++ | ChangeLog.md
Add description for $*USAGE
19:00
pmurias how do I get better execptions from nqp-j, I loose the java land part of things like NullPointerException 20:00
MasterDuke pmurias: are these lines needed? github.com/perl6/nqp/blob/truffle/...nqp#L7-L11 20:29
pmurias MasterDuke: passing the @ARGS is needed 21:14
MasterDuke pmurias: but not the `:module-path('gen/js/stage2')`, etc?
pmurias not that 21:15
MasterDuke: the whole thing is hastily cargo culted from a js cross compiling script, it's not how things should be after we bootstrap, ditch the jvm bytecode backend and run fully on truffle 21:16
I think we will uncommented the optimization insertion soon 21:18
I would need sleep to think over what we can strip from that but not sure how important cleaning that up is 21:20
MasterDuke: is any of that causing problems?
MasterDuke pmurias: nope, just curious 21:21
pmurias ok ;)
it's a temporary hack so I just cargo culted something up at the start of the backend and didn't touch it since 21:22
;)
MasterDuke pmurias: ah, another probably-not-terribly-serious question. some files have a copyright statement at the top, but not all. is it needed?
pmurias: and re temporary hacks and cargo culting, i certainly don't hold any high ground, i'm not sure i've contributed an original line to the truffle stuff, it's pretty much all been copied from you or the original jvm code 21:24
pmurias I'm sometimes copying/adapting stuff from the example simplelanguage/graal.js and I copy the copyright headers too 21:25
I'm not a lawyer but I would prefer to stay as safe as possible in regards to copywrite matters with Oracle 21:26
* copyright 21:27
MasterDuke very true
timotimo watching daina pettit's "perl optimization tidbits", it surprises me greatly that in perl5 there's noticable performance differences between "or" and "||" as well as "and" and "&&" 21:35
my first attempt at measuring perl6 with this gives me all four being pretty much the same speed
MFW "for ^10_000_000 { $x }" is actually faster than "for ^10_000_000 { }" 21:36
"or is about twice as fast as and" in perl5
perl 5.22, to b exact 21:37
Geth nqp/truffle: f502ee0f18 | (Paweł Murias)++ | 2 files
[truffle] Fix bug in coercion
21:44
nqp/truffle: 457108f8f8 | (Paweł Murias)++ | src/vm/jvm/Truffle.nqp
[truffle] Turn 'associative' and 'positional' QAST::Var scopes into ops
nqp/truffle: 113ffca746 | (Paweł Murias)++ | 6 files
[truffle] Implement nqp::hash and nqp::bindkey, nqp::atkey and nqp::elems for hashes
pmurias timotimo: the whole talk seems like something specific to an implementation without a well defined/separated AST layer 21:45
timotimo but you can already output the optree it builds, right?
pmurias timotimo: you can print the optree 21:49
pmurias timotimo: but it seems to be a lot more oriented towards how it will be run 21:50
timotimo: QAST was afaik designed to run on parrot
timotimo i don't know about that 21:51
jnthn No, QAST was the successor to PAST. PAST was designed to run on Parrot. QAST was a step away from that to something more general. 21:54
MasterDuke pmurias: did you see my comment on github.com/perl6/nqp/commit/f502ee...cd4f65ac5f ? 21:56
pmurias fixes things 22:01
Geth nqp/truffle: eab8219077 | (Paweł Murias)++ | src/vm/jvm/runtime/org/perl6/nqp/truffle/NQPHash.java
[truffle] Delete file
22:03
pmurias shouldn't commit stuff past his bedtime ;)
MasterDuke heh, lots of progress though! 22:04
AlexDaniel Zoffix++ # v6.d changelog 22:07
timotimo perl6.org/~timo/ 23:22
timotimo ^- 32bit windows build of rakudo 2018.06 23:33