01:48 dalek joined
[Coke] jnthn: weird. latest version of the -only-250-times script is now running 253 times. 03:54
05:51 domidumont joined 05:56 domidumont joined 06:37 FROGGS joined 06:47 kjs_ joined 07:16 domidumont joined 07:25 zakharyas joined 07:31 lizmat joined 07:38 kjs_ joined 08:12 vendethiel joined 08:36 lizmat joined 09:24 geekosaur joined 09:27 kjs_ joined
dalek arVM: e3cedc1 | timotimo++ | src/strings/utf16.c:
prevent null deref when calling MVM_string_utf16_encode

it passes NULL for output_size. thanks clang-analyzer!		 09:45
timotimo clang analyzer still not very happy about our MVMROOT macro 09:49
jnthn Maybe some day it'll be fixed to be OK with it :) 10:01
yoleaux2 17 Oct 2016 23:03Z <japhb> jnthn: Sadly, OO::Monitors still fails without 'no precompilation'. When I comment it out and run `perl6 -I lib t/precomp.t` I immediately get: Missing serialize REPR function for REPR SCRef (SCRef)
timotimo it's only the viewer that has problems 10:05
jnthn ah, ok
timotimo worst case if we suspect a problem it shows is legit, we can rewrite the code to use push/pop instead of MVMROOT
right now it treats the whole blocked out part as a single line when it comes to placing bubbles that explain the step-by-step thingie 10:06
10:06 kjs_ joined
timotimo i was hoping it could point out something wrong with my callsite_flags_sso code, but no dice :( 10:08
10:28 dalek joined 10:33 zakharyas joined
arnsholt timotimo: Does it still throw a fit over the uthash stuff too? 10:37
I remember getting a huge mess of macroexpanded code and like 20 steps for some kind of error 10:38
And it doesn't deal very well with macros, I agree
timotimo i didn't look at any reports yet that had UTHASH in them, really 10:40
the newest issue i filed had some uthash code nearby, but it wasn't involved
arnsholt No big deal. I'm just curious =)
timotimo mhm 10:42
coverity is quite confused about trycas, did i mention that?
i wonder how many other structs could benefit from compact storage of buffers they own 10:43
having a single object that owns that set of structs is pretty much a prerequisite, though
like the compunit owning all its callsites
10:44 kjs_ joined
timotimo it'll be another necessary step to give our interpreter compact storage for interned callsites, too, of course 10:46
i expect them to be a bit "warmer" than the non-interned ones anyway
i wonder how to figure out the warmth of any given piece of data 10:48
maybe it'd be best to use cachegrind and see where many cache misses happen that seem unnecessary
but the "unnecessary" thing is something a human would probably have to inspect on a case-by-case basis?
dalek arVM/extract-spesh-alloc: aba07a2 | brrt++ | / (7 files):
Extract spesh allocation from spesh

The `spesh` allocator uses a strategy known as memory-region allocation and is suitable for more contexts than just `spesh`.		 11:08
brrt timotimo jnthn ^ the above is the extraction of the spesh allocator 11:24
it's almost a full copy-pasta job
11:57 lizmat_ joined 12:18 vendethiel joined 12:38 kjs_ joined 13:00 kjs_ joined 14:05 kjs_ joined 14:07 zakharyas joined 14:17 kjs_ joined 15:04 vendethiel joined 15:06 zakharyas joined 17:33 hoelzro_ joined 17:34 stmuk joined, moritz joined
dalek arVM: 0402247 | FROGGS++ | src/io/dirops.c:
Revert "rewrite mkdir_p api, no function changes"

This reverts commit 958a01f5ca2d955bfe1e4b3b2c44bbaffeff2d39. This was meant to noch change the behaviour, though it breaks rakudo in windows. Creating a non-existent folder returns an error but succeeds.		 17:36
17:38 domidumont joined 17:39 zakharyas joined, zostay joined
dalek arVM: 14b63ea | (Jimmy Zhuo)++ | src/io/dirops.c:
Revert "Revert "rewrite mkdir_p api, no function changes""

This reverts commit 04022479330216cfca8361a08b2104020ca8f305.		 18:00
arVM: 404ed26 | (Jimmy Zhuo)++ | src/io/dirops.c:
fix broken mkdir on windows
18:00 JimmyZ joined
JimmyZ FROGGS: ^ I just fixed it on windows 18:01
FROGGS lemme check... 18:02
is that new? 18:03
dirops.c
src\io\dirops.c(38) : warning C4133: 'initializing' : incompatible types - from 'wchar_t *' to 'char *'
src\io\dirops.c(46) : warning C4133: 'function' : incompatible types - from 'wchar_t *' to 'const char *'
JimmyZ: works on my box... 18:06
JimmyZ FROGGS: not new warning 18:09
oh warning
dalek arVM/mkdir: 360ddcd | (Jimmy Zhuo)++ | src/io/dirops.c:
fixed mkdir warning on windows
 18:40
FROGGS JimmyZ++ 18:41
JimmyZ FROGGS: ^^ Could you try this branch to see if the warning goes out and still works or not? thanks
FROGGS hold on
JimmyZ I don't have compiler on windows now :)
FROGGS O.o 18:42
JimmyZ: dirops.obj : error LNK2001: unresolved external symbol PathFileExistsW 18:45
dalek arVM/mkdir: ec7dcb4 | (Jimmy Zhuo)++ | src/io/dirops.c:
add missing header on windows
 18:48
JimmyZ FROGGS: ^^
FROGGS dirops.obj : error LNK2001: unresolved external symbol __imp_PathFileExistsW 18:49
JimmyZ hmm , where to add the linking Shlwapi.lib code? 18:50
FROGGS build/setup.pm? 18:51
dalek arVM/mkdir: 529eb04 | (Jimmy Zhuo)++ | build/setup.pm:
add missing linking lib for windows
 18:58
JimmyZ FROGGS: hopes it build now 18:59
FROGGS trying msvc, then gc
gcc*
dirops.obj : error LNK2001: unresolved external symbol __imp_PathFileExistsW 19:02
after reconfigure... 19:03
damn
sorry
wait
(forgot to pull -.-)
JimmyZ any changes to link? :) 19:09
*chance
FROGGS built under msvc successfully 19:10
but now mkdir fails if the folder exists 19:11
which it did not before
JimmyZ only on windows? 19:14
I test it on linux, it works fine here
FROGGS no, on linux too
wait..
JimmyZ: works on linux, fails on windows 19:16
$ perl6 -Mnqp -e 'my $foo = "foo".IO; say $foo.e; say nqp::mkdir($foo.abspath, 0o777)'
True
/home/froggs/dev/rakudo/foo
C:\rakudo>perl6 -e "my $foo = 'foo'.IO; say $foo.e; say $foo.mkdir" 19:17
True
Failed to create directory 'C:\rakudo\foo' with mode '0o777': Failed to mkdir: 258
dalek arVM/mkdir: 0856608 | (Jimmy Zhuo)++ | / (2 files):
more fix on windows
 19:26
JimmyZ FROGGS: ^^ last try, thanks , if still fails, I should install msvc ... 19:27
19:27 pyrimidi_ joined
FROGGS doing... 19:31
19:35 pyrimidine joined
FROGGS JimmyZ: seems to work on msvc 19:37
testing gcc on windows now
JimmyZ: works on gcc/win also 19:41
JimmyZ FROGGS++, thanks
FROGGS JimmyZ++
dalek arVM: df20321 | (Jimmy Zhuo)++ | src/io/dirops.c:
fix mkdir warning on windows
 19:43
20:27 Ven_ joined 21:19 dalek joined
timotimo any reason to not immediately merge the spesh allocator extraction thing? 23:05

Powered by Cro and the Raku® Programming Language.

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

Thank you!