Perl 6 - the future is here, just unevenly distributed

IRC log for #marpa, 2016-10-07

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

All times shown according to UTC.

Time Nick Message
00:22 idiosyncrat_ joined #marpa
01:48 ilbot3 joined #marpa
01:48 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
03:12 ronsavage joined #marpa
04:07 idiosyncrat_ joined #marpa
04:52 idiosyncrat_ Good night!
05:23 ronsavage joined #marpa
07:08 sirdancealot joined #marpa
10:54 kaare_ joined #marpa
13:18 aredridel joined #marpa
16:14 sirdancealot joined #marpa
18:09 harovali joined #marpa
18:39 sirdancealot joined #marpa
18:55 choroba joined #marpa
18:57 Cheery_ joined #marpa
19:08 Cheery joined #marpa
19:14 sirdancealot joined #marpa
19:14 beaugunderson_ joined #marpa
19:33 ernimril joined #marpa
19:33 btyler joined #marpa
19:38 aredridel joined #marpa
19:43 shadowpaste joined #marpa
19:43 iarna joined #marpa
19:47 sivoais joined #marpa
19:48 harovali joined #marpa
19:49 editable-koo joined #marpa
19:56 Pursuit joined #marpa
20:05 aredridel joined #marpa
20:07 sivoais joined #marpa
20:09 lucs joined #marpa
20:12 VsyachePuz joined #marpa
20:18 sivoais joined #marpa
20:22 KotH joined #marpa
20:34 idiosyncrat_ joined #marpa
20:37 idiosyncrat_ A recent interview w/ Randal Schwartz explains some things about how the Marpa FLOSS Weekly interview came about
20:37 idiosyncrat_ https://www.youtube.com/watch?v=JliSGjD1UCE
20:38 ernimril joined #marpa
20:38 idiosyncrat_ It took a while to come off because I saw a posting (somewhere) from Randal about wouldn't a Marpa segment, which seemed peculiar -- why didn't he just contact me I wondered?
20:39 idiosyncrat_ We've known each other for some time.
20:39 editable-koo joined #marpa
20:39 shadowpaste joined #marpa
20:39 btyler joined #marpa
20:39 KotH joined #marpa
20:39 idiosyncrat_ Turns out, Randal's usually dealing with larger projects, and projects not so centered on a single person.
20:40 idiosyncrat_ Unfortunately, at this point, because of Marpa's mathematical nature, and the complexity of its initial implementation, there's a dependence on me.  I expect this phase to pass, but for new that's the case.
20:42 idiosyncrat_ Anyway, with the better dispersed projects, it can be hard to Randal to identify the right contact, so he "cloud sources" it -- announces there will be a talk on X, and waits to hear from somebody from the X team.
20:43 idiosyncrat_ In the meantime, I was having real Internet connectivity problems, and Randal's show is live, so I needed to get someone else.
20:43 idiosyncrat_ Andrew ("the face of Marpa") Rodland kindly allowed himself to be drafted.
20:45 idiosyncrat_ Obviously nobody was going to take it upon themselves to respond for the Marpa community unless I did ...
20:46 idiosyncrat_ and it took time for me to realize I had to reply personally to Randal's broadcast message and once I did realize that, I needed to enlist Andrew's help first.
20:47 idiosyncrat_ Andrew was the choice, btw, because he'd given Marpa talks before (twice), he's contributed meaninfully to its development and very very importantly ...
20:47 sivoais joined #marpa
20:47 idiosyncrat_ because of time zone and first language fluency reasons.
21:17 idiosyncrat_ joined #marpa
21:33 harovali hi  idiosyncrat_!, first of all I want to thank you, not only for the amiful help you always offer, but also for marpa. I'm so much impressed for the marpa effort. I know it's a lifetime devoted to this.
21:33 harovali s/amiful/aimful/
21:35 harovali I was able to advance a little in my comprehension of marpa, and in the little parser i'm trying to implement
21:35 harovali I still have some issues which now I feel I will be able to catch
21:37 harovali In an anecdotal fashion, I want to share something. Yesterday I was trying to put everything together in the grammar, which was commencing to work, but was being dubbed ambiguous by marpa.
21:45 harovali So I tried things to tackle the ambuguity
21:45 harovali I wanted to be more specific in some 'separator' lexemes that I was writing.
21:46 harovali so i added some 'negative match' character classes
21:46 harovali I womdered myself whether another lexeme already defined, could be 'negated'
21:47 harovali so i started typing sort of out of my imagination a character class lik this :  [^:mylexeme:]
21:48 harovali this change made the grammar disambiguate.
21:48 harovali so I started thinkint wonderful things not only about marpa but about me too !
21:49 harovali after some though, I saw that what made the grammar work wasn't that delusional creation of mine, but the fact that something random was being negated
21:50 harovali now , I'm crawling among the gruoping , and the priorization of rules
21:51 harovali because I saw that some of my specialized rules matched, but some later ones did not, and matched in the general case, generating ambiguity, and ultimately rendering the parsing effort not useful
21:52 harovali I still need help, but at least I'm confident of my learning
21:52 harovali Is that capability of regular expression of negating matches but not consuming the negated match, availabe somehow in marpa ?
21:53 harovali is that considered a manual lexing thing ?
22:00 idiosyncrat_ harovali: re https://irclog.perlgeek.de/marpa/2016-10-07#i_13359700 -- yes, exactly, it's deterministic thinking and Marpa is non-deterministic.
22:02 idiosyncrat_ harovali: re https://irclog.perlgeek.de/marpa/2016-10-07#i_13359611 -- thanks.  It has indeed turned out to be my life's work, for better or worse. :-)
22:05 idiosyncrat_ Re negating matches -- you can try other approaches, such as defining lexeme priorities.
22:06 idiosyncrat_ If a lower priority lexeme is ignored in favor of a higher priority lexeme, that is similar in effect to a negated match.
22:27 idiosyncrat_ joined #marpa
22:50 ronsavage joined #marpa
22:58 idiosyncrat_ joined #marpa
23:07 sirdancealot joined #marpa
23:18 harovali idiosyncrat_: I'll take that way
23:19 harovali idiosyncrat_: the ambiguity is a synonim for non-determinism ? If so, why does marpa has a special state equiparable to an error for an ambiguous grammar ?
23:20 harovali ronsavage: I'd also like to thank you for the many instances of help you've given.
23:35 idiosyncrat_ Re ambiguity vs non-determinism -- they are related, but not synonyms.
23:35 idiosyncrat_ Ambiguous means there is more than one parse.
23:36 idiosyncrat_ Non-deterministic means that, in the course of discovering the parse or parses, multiple paths may be pursued at the same time.
23:36 idiosyncrat_ A deterministic parser must always produce an unambiguous parse, if only by ignoring the other alternatives, as in PEG.
23:37 idiosyncrat_ A non-deterministic parser, like Marpa, can produce both ambiguous and unambiguous parses.
23:39 idiosyncrat_ Take the grammar S ::= X | Y ; X ::= a X | x; Y ::= a Y | y
23:39 idiosyncrat_ This (if I got it right) matches a*(x|y)
23:40 idiosyncrat_ That's unambiguous -- the final <x> or <y> always tells you which it is, but it does not come until the end ...
23:41 idiosyncrat_ so Marpa has to keep track of both parses all the way until the last character ...
23:41 idiosyncrat_ which it can do, and in linear time.
23:42 idiosyncrat_ To do that, it uses non-determinism -- it just follows both possibilities, and only selects which is the right one at the end when it sees an <x> or a <y>.
23:42 idiosyncrat_ So non-deterministic, but unambiguous.
23:51 harovali thanks idiosyncrat_ for the detailed expanation
23:51 harovali explanation

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