Welcome to the main channel on the development of MoarVM, a virtual machine for NQP and Rakudo (moarvm.org). This channel is being logged for historical purposes.
Set by lizmat on 24 May 2021.
Woodi right, not everything is struct :) 04:27
Woodi then memory areas for lexical scope would help to track usage. and when something is ascaping pack it into struct ? :) 04:29
Woodi I think my problem with gc's is that looks that programs need to look for things that, somehow, are lost... then mark it and search in circles... my naive wishiness wants to have everything deterministic, black and white - used or not - from parse time... 04:49
MasterDuke timo: nice idea about doing a gc during blinking/saccades. not that applicable for most raku programs probably, but could help make using gc'ed languages more viable for user-interactive things 14:08
and speaking of gc, there was some discussion at the RSC about making the unmanaged size of strings, arrays, etc pressure the nursery, like big integers do 14:09
oh, also done when allocating buffers for io it looks like 14:13
anybody have any thoughts about whether it should be done at all? and then if so, what heuristics to apply? 14:14
oh, s/RSC/RCS/
lizmat hehe :-) 14:30
MasterDuke huh, github.com/MoarVM/MoarVM/commit/bb7e757265 18:12
lizmat maybe timo remembers 18:13
MasterDuke oh, this is almost exactly the same as something else that was discussed colabti.org/irclogger/irclogger_lo...07-19#l347 18:16
lots of heruistics to bikeshed in this area. multiple stands to reduce memory use? fewer strands to make iterating over a string faster? 18:17
lizmat you mean irclogs.raku.org/moarvm/2017-07-19...22:40-0001 ? :-) 18:18
MasterDuke heh. different computers have different tabs already open... 18:27
ugexe so i think we just need to have a github.com/openjdk/jdk/blob/b94b67...#L377-L397 in MoarVM and to use it on any file paths for windows users to fix long paths. all the other backends already just support them afaict 18:32
lizmat would that imply that we could get rid of all the backslashy checks in Raku ? 18:37
aka IO::Spec::Win32 ? 18:38
ugexe Unfortunately no. In fact we have to be extra sure we replace / with \. The reason \\?\ allows long paths on windows is because it bypasses the windows internal path normalization that uses MAX_PATH as one of its parameters. However, that normalization is what allows / in paths to work on typical windows paths. In other words: a \\?\ prefixed path won't work with / like it usually would 21:01
brrt \o 22:01
link drop brrt-to-the-future.blogspot.com/20...m-jit.html 22:33
timo brrt++ 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!