Perl 6 - the future is here, just unevenly distributed

IRC log for #marpa, 2014-05-16

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

All times shown according to UTC.

Time Nick Message
16:24 jeffreykegler joined #marpa
16:25 jeffreykegler My time these days is mainly taken up by a rewrite of the Marpa paper.
16:25 jeffreykegler It's taking more time than I had planned, but I am hoping it is worth it.
16:26 jeffreykegler The new paper will feature proofs "from scratch" of the claims for Marpa -- the previous one builds on those in the literature.
16:27 jeffreykegler It will contain my new result that Marpa (and Leo 1991) is linear for finite unions of LR-regular grammars.
16:28 jeffreykegler And it will also contain a series of theorems that better characterize the grammars for which Marpa is O(n) -- linear.
16:31 jeffreykegler Among unambiguous grammars, Marpa is linear for every grammar except unmarked middle recursions.  "Unmarked" here is not sharply defined, although in practice "you know it when you see it".
16:31 jeffreykegler The paper is an important part of being "open source" in a real sense.
16:32 jeffreykegler There are parsers out there which make claims to be fast, without saying what times they achieve with what grammars, some in the academic language.
16:33 jeffreykegler But the better tradition is to make specific claims, and supply proofs.
16:33 jeffreykegler This not only makes clear to the user what they can expect but, even more important ...
16:34 jeffreykegler the proofs say *why* Marpa works and *why* it achieves the speeds that it does.
16:34 jeffreykegler They are an important supplement to the open source code, which in a sense is not fully open source without them.

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