|
02:10
coleman left
02:12
coleman joined
02:14
guifa joined
03:05
guifa left
10:11
[Coke] left
10:57
[Coke] joined
11:30
sena_kun joined
15:30
bartolin left
|
|||
| lizmat | Sometimes I don't understand why things don't work on Windows | 15:36 | |
| doesn't precomp on raku.land/zef:lizmat/Text::Emoji | 15:39 | ||
| raku.land/zef:lizmat/Text::Emoji doesn't precomp on Windows | 15:40 | ||
| it has 2 large constant lookup hashes | |||
| if I remove "constanr" from them, it works | 15:42 | ||
| I don't see how having something done at compile time should matter whether it is on Windows or not | |||
| apart from the C-compiler used | 15:52 | ||
| hence me posting this find here | |||
| ugexe | i wouldn't expect raku 'constant' to have any connection to C at all. anyway maybe its worth writing a small one-liner that generates a huge hash and does a similar lookup (similar to some of the out of the ordinary performance examples i've been posting to issues lately) | 16:25 | |
| i guess writing a 1 onliner is a bit more difficult since it would need to generate a module and then run a process that uses that module | 16:26 | ||
| lizmat | yeah, it only appears to happen during precomp afaics | 16:27 | |
| do you have another theory as to why this wouldn't work on Windows? | |||
| any other variables that I'm missing ? | 16:28 | ||
| actually, the Text::Emoji dist has a generator script in the tools dir | 16:29 | ||
| the other reason I just thought of, is some kind of issue in grapheme handling that shows up on Windows? | |||
| ugexe | i'd probably still try to test if it is related to sizes somehow, i.e. if a hash of similar size but without emoji also crashes it | 16:32 | |
| timo | it just occured to me that we should take the raku ops .so stuff and put it directly into the runners we are already compiling anyway | 17:17 | |
| need to double-check if the "rakudo early build" thingie can also work without a .so | 17:18 | ||
|
17:22
bartolin joined
22:52
sena_kun left
|
|||