github.com/moarvm/moarvm | IRC logs at colabti.org/irclogger/irclogger_logs/moarvm
Set by AlexDaniel on 12 June 2018.
Kaiepi is Configure.pl no longer supposed to have its executable bit set? 02:20
nwc10 "requires embedding (which iiuc ..." that was the bit I wasn't getting 06:31
AlexDaniel Kaiepi: that doesn't sound right 10:46
Kaiepi: where is that? 10:47
AlexDaniel I see 10:53
Geth_ MoarVM: 8efd382513 | (Aleks-Daniel Jakimenko-Aleksejev)++ | 2 files
Make scripts executable
AlexDaniel Kaiepi: thanks
nine For those curious: the Perl 5 IO::Handle threading issues I came across are caused by perl5.git.perl.org/perl.git/commit/986a805c4b2 This is not thread safe: PL_check[OP_LINESEQ] = io_ck_lineseq; According to perlapi(1) "PL_check" is global to an entire process and must not be written to directly. Instead there's wrap_op_checker 12:30
timotimo oh damn 12:35
nine So replacing that direct array access with wrap_op_checker is half of the fix. The other will be to add some checks to the new op_checker so it won't indiscriminately modify any OP tree it encounters, but only the functions it was meant for 12:39
brrt \o 14:16
dogbert17 hello, brrt 14:38
brrt ohai dogbert17 14:45
I haven't debugged the sp_bind failure yet
dogbert17 do you have any theories as to what might be wrong? 15:33
brrt no 16:01
:-)
making matters worse, rakudo is apparently sensitive to the number of env vars
MasterDuke sensitive in what way? 16:02
timotimo that was one point in the talk that also had coz (the causal profiler) in it 16:04
that changing the amount of commandline flags or environment variables may be enough to disturb the memory layout of you prrogram to change performance significantly 16:05
but that was only for performance
brrt sensitive, in that the exact frame that is compiled by spesh/jit at a specific point in the program, tends to depend on a myriad of factors, even if MVM_SPESH_BLOCKING is set; and that the number of env vars seems to be a significant parameter in that 16:08
.... sometimes I hate rakudo for that
(part of the torment on implementors I'm sure)
so, here's the thing though.
I think that scripting is all about the interface between the script and the thing that it is scripting, i.e. some sort of opaque computational systme 16:09
R, for instance, is essentially a scripting language for a linear algebra library written in FORTRAN 16:10
lua is a scripting language for arbitrary programs that adhere to the C ABI 16:11
perl5 is a scripting language for the POSIX abi, plus a regex engine, plus any libraries loaded via XS, etc
you can go on and on with this
If that is so, and *if* perl6/raku is supposed to be a scripting language, then it doesn't make much sense to *also* have everything out in the open 16:12
having every thing about the compiler/runtime out in the open is not a scripting language property
it is rather a property we see in languages like smalltalk and lisp 16:13
I don't know how to classify that type of language
timotimo can you clarify "out in the open" a bit more? like the ability to use slangs for example? 16:14
brrt for instance, yes
in raku/perl6, the interface between the grammar/regex engine, and the rest of the language, is 'open'
in perl5, python, ruby, javascript, etc. it is very much not so 16:15
timotimo when will we be able to just save a rakudo memory image and ship that as our program
brrt scripting languages have an opaque interface between the script and the thing that they are scripting
or I should say, tend to have that
and that's somewhat crucial as it allows implementors to cheat
</rant> 16:16
brrt is going to dinner
timotimo cdn.discordapp.com/attachments/538...nknown.png - "all types/frames" view shows size and count for stuff that was in the highscore list, later it'll ask for the rest to be filled, but first i want to give a "find N of those" a shot 22:21

Powered by Cro and the Raku® Programming Language.

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

Thank you!