|
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. |
|||
|
00:25
guifa_ joined,
guifa_ left
08:39
lizmat left
15:31
rakkable left
15:32
rakkable joined
18:28
lizmat joined
18:30
lizmat left,
lizmat_ joined
20:16
lizmat_ left
20:17
lizmat joined
|
|||
| patrickb | It's pretty bad to couple moar code to HLL classes, right? | 20:41 | |
| Currently my var-arg approach encompasses passing a named arg in the actual native function call with a `Signature` object as value. That object I now need to inspect in moar land to extract the parameter types. Is there a good alternative? | 20:44 | ||
| timo | maybe that part can be done in a dispatcher. if there were a dispatcher target that is like "invoke native call, every other argument is a type to use for the argument that comes after it" or so, that could be dispatched to after setting stuff up from inside hll code | 21:24 | |
| or the arguments come after all the types or so | |||
| patrickb | the dispatcher idea could maybe work | 21:31 | |
| The reason I want to use an actual Signature object instead of just a list of types in the HLL call is that other wise I have to touch the raku grammar, because "Str is copy" is not legal raku code outside of a Signature literal. | 21:32 | ||
| So the call would look as follows: sub var_fun(Str) is native is variadic {*}; var_fun("val", 1, "two", :var-sig(:(int32 is rw, Str))); | 21:35 | ||
| That's kind of ugly, but pretty consistent I'd say. | 21:36 | ||
| The syntax could be improved (e.g. putting the Signature in as a positional instead: `var_fun("val", :(int32 is rw, Str), 1, "two");`), but the fundamental issue of how to reach into that Signature stays. | 21:38 | ||
| I guess I'll dig a bit into dispatchers. Oh man, I once again underestimated how a seemingly minor project can escalate into a deeeep rabbit hole... | 21:41 | ||
| timo: Thanks for the pointers! | |||
| timo | YW | 21:42 | |
| japhb | patrickb: But isn't the rabbit hole half the fun? :-D | 22:17 | |
| timo | big sigh, turning on coverage logging makes core setting compilation no longer work ... that's amazing ... | 22:35 | |
|
22:46
lizmat left
|
|||
| timo | somehow the `return if $!composed` in Attribute.rakumod is hitting "trying to return outside of any routine" and i don't really understand why | 22:52 | |