Perl 6 - the future is here, just unevenly distributed

IRC log for #marpa, 2017-01-08

| Channels | #marpa index | Today | | Search | Google Search | Plain-Text | summary

All times shown according to UTC.

Time Nick Message
09:15 idiosyncrat_ Good night!
09:16 choroba joined #marpa
16:04 jdurand joined #marpa
16:05 jdurand Re https://irclog.perlgeek.de/marpa/2017-01-01#i_13830213 - Jeffrey is right aking which regexp implementation first. I choose PCRE2 because it has two very nice features: this is a priori as powerful as Perl itself, and it has the PARTIAL_MATCH facility, that allows to work in streaming mode
16:06 jdurand The full implementation is explained at https://github.com/jddurand/c-marpaESLIF/tree/master/doc/BNF
16:07 jdurand This also allows to implement natively lookahead
16:08 jdurand Jeffrey, I have two questions:
16:08 jdurand * Is it possible to ask what would be the expected terminals /if/ an alternative would be pushed ? A sort of terminal prediction lookahead at the grammar level.
16:10 jdurand * In libmarpa there is the untested method marpa_r_clean() that is of interest to me. Also I can probably bypass it with some trickery, I would like to know if this is usuable and what it does exactly. A rollback to the previous step?
16:12 jdurand About regexes again, one would question about its performance. Again, this is totally dependent on the regex library, and PCRE2 has a JIT facility, that makes it very fast.
16:23 jdurand Jeffrey, about marpa_r_clean() I found https://irclog.perlgeek.de/marpa/2016-06-18/text where you already said this was a "projected" method. Therefore, let's skip it, I would like to know if the first question has a reasonnable answer: what would be the next expected terminals if one would push alternatives, though without commiting them...
21:43 ceridwen_ joined #marpa

| Channels | #marpa index | Today | | Search | Google Search | Plain-Text | summary