github.com/moarvm/moarvm | IRC logs at colabti.org/irclogger/irclogger_logs/moarvm
Set by AlexDaniel on 12 June 2018.
lizmat what would be required to create a CCLASS_NEWLINE like nqp::const for the 'Canonical_Combining_Class" property ? 12:17
so that one could easily see if any characters in a string have that property ? 12:18
nine Looks like not much. Just adding the new constant, and implementing the test in MVM_string_grapheme_is_cclass 12:22
lizmat hmmm... 12:40
actually, that's a property that returns strings, I just realized 12:41
so I'd want to check if it returns "0"
but that should be easy, no ?
looking at making Str.raku about 2x as fast 12:43
nine timotimo: is the profiler's heap analyzer already meant to be used? If yes, you need to push your model additions to p6-app-moarvm-heapanalyzer. 13:48
nine Seems like the plain old moarvm-ha doesn't work anymore either :/ 14:04
expected the collectables header at 0xD957A1A6, but got "3\0\0\0" instead.
samcv lizmat, yes that would be pretty easy 16:39
lizmat, so you are trying to check if a string contains "\n" as fast as possible? why doesn't index work? 16:41
timotimo nine: sorry, you'll need to grab the heapanalyzer from my github account, or set the moarvm heap snapshot format version number with a define 16:42
samcv ah. you want to check a canonical combining class? which specific property for what reason do you want to check?
a nqp::index except it works with a propcode and propval would make sense to implement 16:50
nine timotimo: both format 2 and 3 need changes in p6-app-moarvm-heapanalyzer. 16:52
timotimo: is there a special branch? I did look in your github account, but only found commits from 2016 for the heapanalyzer 16:53
timotimo oh, was i working in a branch? that might be the case 16:59
i'm on nqp-ops-makes-stuff-faster at that moment
sorry about the mess 17:00
lizmat samcv: the goal was to speed up Str.raku (formerly known as Str.perl) 18:10
there's a whole loop for looking up special cases. Which could be evaded if we had a fast way of making sure there are no special cases in a string
samcv: on second thought: it propably doesn't make any sense anyway, because we need to check for specific exceptions anyway 18:35
samcv lizmat, the fact uniprop-int does not return 0 for "0" is a bug. not sure how much impact it has on things, though it was on my bucket list 22:08
lizmat it was using uniprop-str 22:46
samcv lizmat, yeah i looked at the code. thanks for making it faster though :) 23:01