Perl 6 - the future is here, just unevenly distributed

IRC log for #marpa, 2015-11-13

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

All times shown according to UTC.

Time Nick Message
00:34 Cheery jeffrey might like to know.. I'm studying his paper in order to implement it into my programming language, as a standard library.
00:34 Cheery as a script or binary.
00:35 Cheery once I looked at my version of earley parser, I sort of concluded that I may want to reuse libmarpa.
00:37 Cheery but then the libmarpa is quite massive.
00:37 Cheery and the paper's actually quite good.
00:38 Cheery I'm at the ayspool horsecock -page.
01:42 Cheery ok.
01:42 Cheery finished reading all the papers
01:42 Cheery getting some sleep over it.. then I'll reflect this on the stuff I originally wrote and see whether I come up with something more.
01:49 Cheery oh, and in case someone cares: https://gist.github.com/ch​eery/bfb249b1f9a327d150f8
01:50 Cheery it's right out of your thesis. you may want to verify I understood it right.
01:52 Cheery oh awesome.. I could vastly improve the results of the earlier parser, if I do couple things right.
01:57 Cheery hm.
01:58 Cheery but then I practically have ended up to marpa. The thing I made is actually few pieces short of marpa.
02:24 Idiosyncrat Cheery: You may want to know
02:25 Idiosyncrat 1.) I reimplemented Libmarpa some time ago to not use the Aycock-Horspool LR(0) states -- they got in the way of new features, like events ...
02:25 Idiosyncrat and contributed almost nothing in terms of performance.
02:26 Idiosyncrat 2.) Because of this, and because my previous theory paper is attracting readers, I am rewriting it.
02:27 Idiosyncrat Note that the Aycock-Horspool influence was not entirely eliminated -- figuring out how to deal with nullables have bedeviled Earley parsing since Earley's first papers.
02:27 Idiosyncrat And IMHO their work points the way to the solution.
02:41 Cheery joined #marpa
02:48 ilbot3 joined #marpa
02:48 Topic for #marpa is now Start here: http://savage.net.au/Marpa.html - Pastebin: http://scsys.co.uk:8002/marpa - Jeffrey's Marpa site: http://jeffreykegler.github.io/Marpa-web-site/ - IRC log: http://irclog.perlgeek.de/marpa/today
04:33 ceridwen joined #marpa
04:33 ceridwen joined #marpa
05:22 ronsavage joined #marpa
11:30 Cheery marpa has really changed how I think about parsing, or designing programming languagse.
11:31 Cheery it reduces question of syntax into disgusting politics
11:31 Cheery or user preference when you embed it into your programming language
11:32 Cheery text editing stopped being 'legacy' in any sense of the word
11:33 Cheery because I know that I can write grammars that can be read by the editor.
14:15 Cheery arghl.. almost managed to handle it from paper.
14:15 Cheery but I'm wondering about the memoize_transitions -step
14:17 Cheery how do I form the LIM?
14:37 Cheery what does: "postdot transition unique by rule" mean?
16:35 Idiosyncrat Cheery: The "Leo uniqueness requirement" is explained more in Section 7, p. 12.
16:36 Idiosyncrat For a Leo memo to be created its symbol must be "postdot-unique" at its location.
16:36 Idiosyncrat That is, if the symbol is A and the Leo memo is to be created at location j.
16:37 Idiosyncrat At most one of the EIM's at location j must have A as its postdot symbol.
16:37 Idiosyncrat EIM = "Earley item"
16:38 Idiosyncrat All of this is what I'm rewriting at the moment, ...
16:39 Idiosyncrat so in a few days I'll produce a new, more thorough explanation.
16:39 Idiosyncrat The new one will be very math-y, but explained from 1st principles, so one can work one's way through, ...
16:39 Idiosyncrat starting with only a minimal knowledge of parsing theory.
16:43 Idiosyncrat https://twitter.com/Amsterdam​Xpm/status/665092403255222272
16:45 Idiosyncrat Going back to _user2's question on Perl 6 the other day: http://irclog.perlgeek.de/m​arpa/2015-11-03#i_11475860
16:46 Idiosyncrat Larry's point in the tweet is interesting -- the Perl's are unusual in that their *goal* is unique.
16:46 Idiosyncrat Most languages compete with other languages which are trying to do the same thing.
16:47 Idiosyncrat Only Larry has the goal of being "postmodern", which is sort of a non-goal goal.
16:48 Idiosyncrat Lua also has a unique goal, btw -- AFAIK Lua is the only language trying to be severely minimal -- sort of the opposite goal to Larry's.
16:49 Idiosyncrat As long as Larry is the only one trying in his particular direction, his new efforts are bound to have *some* interest.
16:49 Idiosyncrat Especially if you take into account his record of success.
16:50 Idiosyncrat They say "once you're lucky, twice you're good", but that doesn't apply to creating programming languges.
16:50 djns joined #marpa
16:50 Idiosyncrat You don't luck into something like Perl 5
16:52 Idiosyncrat Cheery: back to "postdot transition unique by rule"
16:53 Idiosyncrat The "by rule" part has to do with the LR(0) states, which Libmarpa no longer uses in EIM's, and which I recommend you don't use -- that is, instead
16:53 Idiosyncrat of an LR(0) state, use a dotted rule, like a tradition Earley item.
16:54 Idiosyncrat Anyway, if you do use LR(0) states, as I was doing when I wrote that original paper, you can have the same dotted rule in two different LR(0) states.
16:55 Idiosyncrat So "postdot transition unique by rule" means that the postdot transition has to be unique in terms of rules, which is not necessarily the ame as being unique in terms of LR(0) states.
16:55 Idiosyncrat Complications like this, which arose all over the place and got in the way of new features ...
16:56 Idiosyncrat were the reason I stopped using the LR(0) states.
18:09 ernimril will be interesting to read the paper when the updates about this is done
22:31 ronsavage joined #marpa
23:16 Idiosyncrat joined #marpa
23:26 ronsavage joined #marpa

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