dalek kudo/lexical_module_load: 3b107d8 | niner++ | src/Perl6/World.nqp:
Fixup "Make importing globals from loaded modules lexical"

Remove commented code		 07:35
kudo/language_versions: dd83283 | niner++ | / (8 files):
First attempt at adding a CORE.d setting
 07:46
kudo/language_versions: c38842c | niner++ | / (7 files):
Bring back IO::CatPath and IO::CatHandle in 6.d

nine@sphinx:~> perl6 -e 'IO::ArgFiles.new;' nine@sphinx:~> perl6 -e 'use v6.d; IO::ArgFiles.new;' Could not find symbol '&ArgFiles'
nine Rebased language_versions on current nom for easier playing with it.
gfldex `Proc::Async.new(|<find / -iname '*.txt' -print>);` trips over malformed utf8 filenames and the exception cant be caught. Is that a known problem? 09:22
dalek kudo/nom: cd765e6 | lizmat++ | src/core/Shaped1Array.pm:
Mark 1dimmed.reverse/rotate as nodal

Because List.reverse/rotate also are		 09:42
lizmat afk&
dalek p: 1b88060 | (Pawel Murias)++ | src/vm/js/Operations.nqp:
[js] Use :side-effects instead of the incorrectly spelled :sideffects
 10:00
jnthn gfldex: Are you using a fairly recent (last month or so) Rakudo? The uncatchable encoding exception thing was fixed not so long ago (along with making it possible to use :enc on Proc::Async) 10:09
S17-procasync/encoding.t covers these, fwiw 10:11
gfldex jnthn: build from today. I will investigate further. 10:20
jnthn: may also be a general io-threading problem 10:27
jnthn: it's not a heisenbug. Here is the example and the error message. I did expect at lease one of the CATCH blocks to catch. gist.github.com/gfldex/84af33884d3...225e474cc9 10:38
jnthn Oh 10:42
No, none of those will
The top-level one is not in the dynamic scope of whatever thread happens to handle the I/O
And you never make it into the tap "we got a message" callback if there's an exception 10:43
You can do something like: $find.stdout.tap(-> $line { print $line }, quit => { say .^name });
DrForr For those of you not on FB/Twitter: I've been accepted to OSCON Austin 2017 and Open Source Days in Copenhagen. 10:45
nine DrForr: \o/ 10:47
DrForr Dunno why it's plural days when it's just one day of conferences. Probably they're assuming more days of beer. Ne'er been to Copenhagen, can't wait to see. 10:49
jnthn Copenhagen is nice. Has some really nice (though expensive) beer. 10:50
DrForr Doubly looking forward to it. 10:51
gfldex jnthn: that works, thanks 10:53
jnthn The Mikkeller bar is worth checking out too fwiw
s/too//
Also, "one of the", they have two
Or did...maybe there's more now :)
gfldex: Also consider using supply/react blocks. You get a QUIT phaser for handling async exceptions. 10:54
Not to mention subscription management.
DrForr jnthn++ # suggestion added. 10:58
nine Thinking about the language version vs. compiler version discussion: the wish for depending on compiler versions comes from this version being a proxy for "this feature or bugfix I need is implemented". 11:25
|Tux| This is Rakudo version 2016.10-283-gcd765e6 built on MoarVM version 2016.10-44-g4c9fd00
csv-ip5xs 2.984
test 13.725
test-t 6.373
csv-parser 14.612
nine That reminds me of autoconf. Not that I wish that particular hell upon us, but the principle to have configure time checks for features sounds entirely sane in a world of multiple different compilers and runtimes. 11:26
gfldex jnthn: react works nicely, I do get the expection. However, there seams to be now way to avoid the malformed utf-8 problem. I get X::AdHoc+{X::Promise::Broken} on calling Buf.decode and I'm not able to handle the exception and thus continue with the next line from find. That makes Proc::Async pretty useless. 11:30
pmurias nine: turning a compiler version requirement in a configure time check would require figuring out why exactly some old version doesn't work 11:33
nine pmurias: which requires nailing down that it's actually really a compiler bug and not some coincidence that my bug is exposed on compiler version A but not B. That may actually be a good thing. 11:34
Thinking further, we add new spec tests when we fix bugs in rakudo which would be handy for those configure time checks. But new spec tests mean we create a new version of the spec test suite which _makes it_ a new language version. 11:36
As a module author you essentially want to say "I need a compiler that passes this version of the spec test suite" and right now we only have a single official version and lots of git hashes... 11:37
Long story short, v6.c.1 seems to be long overdue. 11:38
jnthn gfldex: Just use a different encoding 11:42
timotimo utf8-c8, you mean? 11:43
except it's still buggy :)
jnthn That'd do it, yes
Has corner cases, true
gfldex now i'm getting `X::AdHoc Unknown system error 0` 11:45
timotimo error 0? that doesn't sound like an error :\ 11:46
jnthn has no idea what that is about
gfldex gist.github.com/a4d6908a2f1072bdcb...f88b8f8f82 11:48
happens every time but on different spots while dealing with stdout from find 11:49
it's a reliable heisenbug :)
is there a list of provided encodings? (that's ENODOC) 11:50
jnthn Can reproduce it here too 11:51
Please RT; can look at that tomorrow
gfldex .backtrace is empty 11:54
jnthn .backtrace.full maybe? 11:55
timotimo async_spawn_stdout_bytes_read is calling async_read with an nread of a negative value that's not UV_EOF
actually, nread is 0 11:56
ah
we're apparently treating "0 bytes read because we're waiting for some more data" as an error condition
gfldex .backtrace.WHAT is Str and thus no .full 11:58
is that intentional or did I found my 2nd bug today? 11:59
dalek kudo/nom: c5d33a2 | lizmat++ | src/core/Shaped (2 files):
.STORE should return self

Unfortunately, this doesn't fix the problem of .rotate/.reverse on 1dimmed shaped arrays		 13:07
timotimo not entirely sure whether the fix is to handle an nread of 0 in async_read or not to call async_read if nread is 0 13:18
dalek ast: f1581ac | samcv++ | S02-literals/quoting.t:
Add tests for Qw, Qx and Qww. One more test for qx to check how two backslashes are interpreted (#182)

  * Add tests for Qw, Qx and Qww. One more test for qx to check how two backslashes are interpreted
  * Compare that the result of Qx[echo '\'] and qx[echo '\\'] are equal so it works on Windows as well as POSIX shell
  * Compare \\ and \\\\ not \ and \\
  * Compare Qx[echo '\\\\'] and qx[echo '\\\\\\\\'] not \\ and \\\\		 13:54
ast: 8b2316f | lizmat++ | S03-metaops/reverse.t:
Fix typo in comment
 14:30
gfldex jnthn: as requested: RT#130107 16:08
synopsebot6 Link: rt.perl.org/rt3//Public/Bug/Displa...?id=130107
[Coke] RT #129909: "This breaks my heart, since I was planning on delivering a compiled binary to my clients" 16:40
synopsebot6 Link: rt.perl.org/rt3//Public/Bug/Displa...?id=129909
[Coke] Do we have any nicely phrased text that suggests perhaps setting up a donation or something to TPF would help in this regard. (And would it help in this particular case?) 16:41
jnthn gfldex: Will try and fix that up tomorrow 16:47
gfldex "it was fixed within 24h" is always a good line in any blog post :) 16:48
mst wonder how amenable rakudo would be to building something like schmorp's staticperl 17:14
jnthn mst: Is that "all libraries/scripts, VM, runtime etc. bundled into a single static executable"? 17:21
Or am I confusing it with another thing?
mst jnthn: yes. 17:33
dalek p: a900659 | FROGGS++ | tools/build/MOAR_REVISION:
bump moar for clock_gettime issue on macOS
 18:11
kudo/nom: ee8ae92 | FROGGS++ | tools/build/NQP_REVISION:
bump nqp/moar for clock_gettime issue on macOS
 18:12
FROGGS o/ 18:15
masak software.schmorp.de/pkg/App-Staticperl.html -- innerestin' 18:28
FROGGS masak: I used it in the past 18:39
masak I didn't know about it; I picked it up from mst and jnthn's discussion above 18:40
FROGGS masak: the linux binary here was built with staticperl: gamoscope.com/ 18:55
binaries, even
masak cool 19:01
dalek ast: 2001b37 | usev6++ | S32-str/parse-base.t:
Fudge failing tests with unicode literals for JVM
 19:52
viki :( 20:04
j: say +"໖"
camelia rakudo-jvm 76b061: OUTPUT«Cannot convert string to number: base-10 number must begin with valid digits or '.' in '⏏໖' (indicated by ⏏)␤ in block <unit> at <tmp> line 1␤␤Actually thrown at:␤ in block <unit> at <tmp> line 1␤␤»
dogbert17 is jnthn or perhaps timotimo around by any chance? 20:48
[Coke] there's always .ask
dogbert17 never used that, will try 20:49
[Coke] .ask dogbert17 see?
yoleaux2 [Coke]: I'll pass your message to dogbert17.
dogbert17 aha
yoleaux2 20:49Z <[Coke]> dogbert17: see?
[Coke] plus the channel's logged. usually safe to just throw it out there. 20:50
dogbert17 ok
timotimo hi
dogbert17: ?
dogbert17 this is an oldie but I would like to ask again, it concerns an example from the docs: gist.github.com/dogbert17/67228d08...60ae5b78b9 20:51
[Coke] GAH HE WAS HERE THE WHOLE TIME!
dogbert17 do you recognize this timotimo?
[Coke] dogbert17: that seems to work here.
timotimo what's wrong with it? 20:52
dogbert17 if I change the name of the program to something that does not exist, e.g. hexdump2 the program exists without any message at all, is that correct behaviour?
timotimo oh, that thing
dogbert17 oldie :-)
[Coke] if this is an oldie... is there an RT? 20:53
dogbert17 not from me anyway
dogbert17 hides
but is it a bug? 20:54
timotimo it probably ought to break that start promise immediately
dogbert17 running in gdb gives SIGPIPE
timotimo oh, it gets asploded
[Coke] if there isn't a ticket, odds that it will get fixed go way down.
timotimo i thought it just exits, not crashes
dogbert17 there is no visible asplode with perl6, perl6-gdb-m stops with SIGPIPE 20:55
[Coke] (visible) check "echo $?" 20:56
timotimo oh, maybe the shell doesn't show it as asploded because that's how things die when you pipe them through less or something and quit less before the program finished outputting stuff
[Coke] after running the exploding .p6
I get "141"
geekosaur there's a couple of RTs for that issue with missing commands
[Coke] which is sigpipe, IIRC. 20:57
dogbert17 yes, I also get 141
it would have been nice if some error msg was printed as well 20:58
timotimo we're not supposed to die with sigpipe at all 20:59
geekosaur rt.perl.org/Public/Bug/Display.html?id=125651
[Coke] not saying this should be required, but: 21:00
gist.github.com/coke/814af2f3c5378...69944719b3
dogbert17 'no such file or directory' sounds like a nice message
what does 'bindexpayload needs a VMException, got P6opaque (Scalar)' mean? 21:01
arnsholt You need to add a decont, sounds like 21:02
dogbert17 that's from [Coke]'s gist
[Coke] oh, for me that says: 21:04
BOOM
broken pipe in block <unit> at a.p6 line 9
This is Rakudo version 2016.10-17-g200364a
dogbert17 I get BOOM 21:05
bindexpayload needs a VMException, got P6opaque (Scalar)
in block <unit> at /home/dogbert/scratch.pl6 line 9
[Coke] perl6 --version?
dogbert17 This is Rakudo version 2016.10-285-gee8ae92 built on MoarVM version 2016.10-55-g20c8591
[Coke] m: gist.github.com/coke/814af2f3c5378...69944719b3
camelia rakudo-moar ee8ae9: OUTPUT«Proc::Async is disallowed in restricted setting␤ in sub restricted at src/RESTRICTED.setting line 1␤ in method new at src/RESTRICTED.setting line 32␤ in block <unit> at <tmp> line 7␤␤»
[Coke] ah well.
dogbert17 BOOM 21:21
bindexpayload needs a VMException, got P6opaque (Scalar)
at gen/moar/m-CORE.setting:27011 (././CORE.setting.moarvm:rethrow)
from gen/moar/m-CORE.setting:38685 (././CORE.setting.moarvm:result)
from gen/moar/m-CORE.setting:40747 (././CORE.setting.moarvm:await)
FWIW 21:22

Powered by Cro and the Raku® Programming Language.

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

Thank you!