Tagged Deterministic Finite Automata with Lookahead · HackerPapers discussions

Tagged Deterministic Finite Automata with Lookahead — mentioned by chubot

Thu, 22 Feb 2024 16:22:52 GMT

Hm as an re2c user, I always thought that Ragel vs. re2c was a push/pull issue (though re2c flag has a --storable-state flag to generate the push style -- I haven't used it) But now that I look at the example on the front page: http://www.colm.net/open-source/ragel/ action dgt { printf("DGT: %c\n", fc); } action dec { printf("DEC: .\n"); } number = ( [0-9]+ $dgt ( '.' @dec [0-9]+ $dgt )? I don't like how the dgt and dec actions are "inline" in the regex. This seems to be breaking the key abstraction of regexes, which in my mind is "NON-DETERMINISM for free". The better mechanism is submatch…

Paper and discussion history. A mention is not an endorsement.

Tagged Deterministic Finite Automata with Lookahead — mentioned by chubot

Tue, 30 Jun 2020 19:54:30 GMT

Hm I don't really think RE2 is even that influential (unfortunately IMO). Rust is the only new language whose engine borrows its design AFAIK. For example, I was disappointed to learn that Julia uses PCRE. ---- I think backtracking is firmly in the past because it causes too many problems in practice [1], but I think there's scope for a text matching engine that's built on top of regular languages, but more expressive. As mentioned here and on lobste.rs recently, regular languages aren't expressive enough for syntax highlighting (or lexing most programming languages in their entirety). But it…

Paper and discussion history. A mention is not an endorsement.

Tagged Deterministic Finite Automata with Lookahead — mentioned by chubot

Sat, 27 Jun 2020 23:22:13 GMT

OK thanks for the clarifications. I responded to BurntSushi and others here: https://lobste.rs/s/fq8uil/aho_corasick https://news.ycombinator.com/item?id=23665341 I don't have experience with captures in re2c, but it's entirely automata-based, and they apparently implement something "fast" for captures and published it in this 2017 paper, so it's pretty new: https://arxiv.org/abs/1907.08837 I'm sort of interested in benchmarking some Sublime-ish workloads with it, but I'm not sure I'll have time. ---- And the other claim from that comment is that while regular languages aren't powerful enough…

Paper and discussion history. A mention is not an endorsement.