00:11
gfldex left
00:22
lizmat joined
00:32
gfldex joined
00:40
gfldex left
00:42
gfldex joined
01:30
Manifest0 left
06:00
discord-raku-bot left
06:01
discord-raku-bot joined
08:24
librasteve_ joined
09:01
Manifest0 joined
09:14
dakkar joined
12:55
Guest98 joined
13:23
Guest98 left
13:33
librasteve_ left
|
|||
renormalist | in context of grammars, just to make sure: regex/token/rule only *behave* differently (like token doesn't backtrack), but all can contain the full syntax of complex regexes/tokens/rules, right? | 17:28 | |
What nags my brain is the impact of such a behaviour in the overall picture. Like a token 'dummy' which does not backtrack but contains "foo"|"bar" as part of another 'regex' and partially succeeds with "foo", will that *regex* backtrack and potentially try the token 'dummy' with "bar"? | 17:35 | ||
17:39
librasteve_ joined
|
|||
librasteve | my freind ChatGPT says this chatgpt.com/share/673e1fe7-2354-80...8ce6654164 (which is a lot better answer than I can do ... but it may be hallucinating) | 17:44 | |
17:47
dakkar left
|
|||
renormalist | librasteve_: if this sentence is true: "token: | 17:51 | |
"Does not backtrack internally" | 17:52 | ||
then the "internally" is probably my answer. | |||
which it what it explains in 3. | |||
renormalist is a bit surprised by the first usefule complex answer from a ChatGPT - if it doesn't tell rubbish, which I hope. | 17:55 | ||
librasteve | yeah - probably needs testing --- I was surprised at how good the answer was also --- my theory is that raku is underdocumented on the web in general so it probbaly only had one relevant source text and didnt try to weave together a bunch of similar but different texts | 17:56 | |
renormalist | That's my explanation, too. | 17:58 | |
so thanks, this helps me further. | 17:59 | ||
antononcube | With relevant prompts, LLMs know Raku very well. | 18:59 | |
21:50
Heptite joined
23:16
teatwo left
23:18
teatwo joined
23:19
teatwo left
23:20
teatwo joined
23:23
teatime joined
23:24
teatwo left
|