Perl 6 - the future is here, just unevenly distributed

IRC log for #marpa, 2015-08-24

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

All times shown according to UTC.

Time Nick Message
00:04 idiosyncrat_ joined #marpa
00:42 idiosyncrat_ If others want to look at lwa's new https://github.com/latk/p5-​MarpaX-Grammar-Preprocessor ...
00:43 idiosyncrat_ I'd be interested in their reactions.
00:43 idiosyncrat_ I note his SLIF extension/wrapper replace my syntax for adverbs:
00:43 idiosyncrat_ keyword => value
00:44 idiosyncrat_ with something a bit TeX-like:
00:44 idiosyncrat_ \adverb
00:44 idiosyncrat_ or
00:44 idiosyncrat_ \adverb value
03:04 mauke_ joined #marpa
03:09 CQ_ joined #marpa
04:02 idiosyncrat_ good night!
04:04 dvxd good night!
04:37 CQ this is weird, when I click on the MarpaX link from idiosyncrat it adds characters: https://github.com/latk/p5-%E2%8​0%8BMarpaX-Grammar-Preprocessor ... hovering over the link shows correctly (chatzilla)
04:37 CQ same if I copy the link and paste in an editor
05:22 CQ anyone else seeing that, or is it on my end?
05:22 CQ or is it the smallest hidden buffer overflow attack ever? :)
05:23 CQ afk
06:08 ronsavage joined #marpa
06:22 CQ_ joined #marpa
07:37 ronsavage joined #marpa
07:39 pczarn joined #marpa
07:58 ronsavage CQ: For me the link in my irc client gives a 404 but in the Chrome-backlog the link text only goes up to 'p5-'. Weird.
08:00 pczarn There's a zero-width space in the middle.
08:33 koo7 joined #marpa
09:04 CQ2 joined #marpa
09:05 koo7 joined #marpa
11:35 CQ2 joined #marpa
12:35 koo7 joined #marpa
12:40 koo7 joined #marpa
14:07 KotH_ joined #marpa
14:07 djns joined #marpa
15:05 koo7 joined #marpa
15:48 idiosyncrat_ joined #marpa
15:49 idiosyncrat_ The link at http://irclog.perlgeek.de/m​arpa/2015-08-24#i_11104373 seems to be screwy
15:50 djns hi
15:51 idiosyncrat_ Trying again https://github.com/latk/p5-​MarpaX-Grammar-Preprocessor
15:52 djns hopefully my ignorant questions about BNFs have not been a nuiscance to you
15:52 idiosyncrat_ djns: certainly not
15:52 idiosyncrat_ I didn't mean to imply that
15:53 djns oh thank you
15:55 idiosyncrat_ One of the things I have to learn now that I am an idiosyncrat is to be careful with jokes ...
15:55 idiosyncrat_ it was never a problem before because nobody took what I said seriously :-)
15:58 idiosyncrat_ lwa: re the name MarpaX::Grammar::Preprocessor
15:59 idiosyncrat_ I wonder, if it that's best -- it's kind of generic and I expect there'll be other approaches to preprocessing.
16:02 djns yes
16:02 idiosyncrat_ Maybe some sort of "trademark" or "brand name", like MarpaX::PPIF or MarpaX::CoolName
16:04 djns right, I was going to call my lexer PLEX until I found out it was already taken
16:07 pczarn joined #marpa
17:47 koo7 joined #marpa
18:07 koo7 joined #marpa
18:44 pczarn Fortunately, the name PARL is available.
18:46 mauke pczarn: https://metacpan.org/pod/release/RSCH​UPP/PAR-Packer-1.026/script/parl.pod
19:01 idiosyncrat_ left #marpa
19:02 idiosyncrat_ joined #marpa
19:02 idiosyncrat_ AFK
19:51 CQ probably known to you guys, but ... https://swtch.com/~rsc/regexp/regexp1.html
20:08 mauke yeah, that thing
20:10 mauke "In contrast, there are no regular expressions that are pathological for the Thompson NFA implementation."
20:11 mauke there's some sleight of hand going on here
20:11 mauke it's easy to avoid pathological regexes if you don't implement 70% of the features
20:13 mauke ((x{9999}){9999}){9999}[abc]  <- try running that on your Thompson NFA and tell me how far you get
20:40 djns Its shocking how poorly perl performs in those examples
20:40 djns true
20:43 pczarn joined #marpa
20:48 btyler_ CQ, djns: there's rather good discussion of that paper on perlmonks: http://perlmonks.org/?node_id=597262 with some commentary from one of the major contributors to the perl regex engine
21:00 ernimril joined #marpa
21:08 ceridwen mauke, I'm pretty sure that's a "regular expression" that doesn't represent a regular language.  While the term "regular expression" is rarely used in its original technical meaning these days by many software engineers, I still think it's fair to say "regular expression" and mean "a particular representation of a regular language."
21:09 mauke how is that not regular?
21:10 ceridwen How do you represent it with a DFA?
21:10 ceridwen Recognize it, I should say.
21:12 mauke easy, just build an automaton with 999700030002 nodes
21:12 mauke or is that states?
21:14 mauke S0 -(x)-> S1 -(x)-> S2 -(x)-> ... -(x)-> S999700029999
21:14 mauke and from there three arrows (a, b, c) to "done"
21:15 ceridwen Okay, fair.  I misread it.
21:17 mauke the point is that in a naive backtracking implementation, bounded repetition can be done with a counting loop
21:17 ernimril joined #marpa
21:17 mauke if you don't do that, you have to deal with state explosion
21:22 ceridwen Yes, that is a problem.  I don't know if the more refined Thompson NFA implementations have conquered it.
21:23 mauke I doubt it
21:58 idiosyncrat_ joined #marpa
22:17 purmou joined #marpa
22:23 ronsavage joined #marpa
22:55 idiosyncrat_ re the Russ Cox regular expression article and Perl regexes
22:55 idiosyncrat_ Perl's tradition is anti-academic, or at least anti-CS academic -- Larry is a linguist.
22:57 idiosyncrat_ Languages from outside the CS academic tradition can transcend the CS herd mentality
22:57 idiosyncrat_ Whatever you think of Perl now, its effect on computer languages was revolutionary.
22:58 idiosyncrat_ R is another language invented by non-CS people, and it is a leader in literate programming.
22:58 idiosyncrat_ So Perl has an anti-CS heritage -- "ignore the textbooks and just hack"
22:59 dvxd newlisp too. :)
22:59 idiosyncrat_ And for all the faults of its implementation, Perl's regexes were also ground-breaking.
23:00 dvxd If Ken Thompson and Chuck Moore sat down for coffee to design a LISP implementation, and Larry Wall was hiding under the table, they would have come up with newLISP.
23:00 idiosyncrat_ But the textbooks do contain useful stuff, and Russ caught them on some stuff.
23:00 dvxd true that; very groundbreaking.  Larry Wall is a humble genius.
23:00 dvxd Also, he reminds me of Ned Flanders.
23:01 idiosyncrat_ Marpa, while it comes out of the Perl community, is often better accepted outside it -- as the composition of this channel suggests.
23:02 idiosyncrat_ And that's because Marpa is very much based on the textbooks --
23:02 idiosyncrat_ it is not one of those "let's sit down and hack out a parser" projects.
23:08 idiosyncrat_ Mark Dominus is the one example of somebody I know whose approach is basically CS-academic, but who somehow manages to be considered mainstream by the Perl community.
23:15 djns My lexer is basically a "lets sit down and write a lexer' thing, though in developing its finite automata and cloning, thought was put into it to make it versatile

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