🦋 Welcome to the IRC channel of the core developers of the Raku Programming Language (raku.org #rakulang). This channel is logged for the purpose of history keeping about its development | evalbot usage: 'm: say 3;' or /msg camelia m: ... | Logs available at irclogs.raku.org/raku-dev/live.html | For MoarVM see #moarvm Set by lizmat on 8 June 2022. |
|||
11:24
MasterDuke joined
|
|||
MasterDuke | i think github.com/rakudo/rakudo/pull/5887 is good to go | 11:28 | |
tellable6 | hey MasterDuke, you have a message: gist.github.com/1ade18f7594e013431...49a5ffb630 | ||
MasterDuke | nine: however, i'm not sure what the rules are for when an (UN)WRAP can be removed. with the PR as is, all the same spectests pass (with RAKUDO_RAKUAST=1) as on main | 11:31 | |
but there are at least some UNWRAPs that i can currently remove, but istr you said shouldn't be, because the function could be overridden in HLL/user code | 11:33 | ||
is a simple heuristic that the return value of every not-all-uppercase method needs to be UNWRAPped correct? | 11:35 | ||
e.g., `my @foo = self.IMPL-UNWRAP-LIST($bar.this-could-be-overridden-and-must-be-UNWRAPped)`, but `@my foo = $bar.NO-NEED-TO-UNWRAP-THIS)`? | 11:37 | ||
11:54
MasterDuke left
|
|||
ab5tract | MasterDuke: how big are the performance gains from these changes? Unless it is significant, I think it’s a lot easier to reason about as is rather than having some instances where unwrapping is necessary and others where it isn’t | 12:26 | |
tellable6 | ab5tract, I'll pass your message to MasterDuke | ||
ab5tract | nine: is the wrapping stuff something that could be addressed in a RakuAST based optimizer? Or is the surface area too opaque to be able to know when wrapping can be skipped? | 13:35 | |
19:02
sivoais left
19:04
sivoais joined
23:17
MasterDuke joined
|
|||
MasterDuke | ab5tract: they're not very significant. but they are a simplification in the sense that they're removing code. fwiw, i'm willing to pay a decently high price for better performance, especially since the rakuast frontend is currently quite a bit slower | 23:20 | |
23:23
MasterDuke left,
MasterDuke joined
23:49
MasterDuke left
|