Perl 6 - the future is here, just unevenly distributed

IRC log for #marpa, 2016-04-20

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

All times shown according to UTC.

Time Nick Message
00:00 ronsavage No comment inside (# ... #). /That/ does not worry me. Alternatives are html-style (vaguely) <- ... -> (I gues you won't like that :-) or an adaptation of the comment such as...
00:01 ronsavage #! ... #! or similar.
00:01 ronsavage Oops. I meant #! ... !# of course. Or what about /* ... */?
00:26 idiosyncrat_ I suspect if Larry Wall were doing the DSL design, he wouldn't have a problem with doing some special lookahead hackery to make comments work inside '(# ... #)' groups, and he'd be happy to deal with the cases the hackery doesn't cover by saying "Just don't do those things".
00:27 idiosyncrat_ But, while Larry studied linguistics, I studied math, and I'm not as non-orthogonality-tolerant as Larry probably would be.
00:33 harovali joined #marpa
01:09 ronsavage Yes, but! Let me ask: Has the comment syntax itself be decided upon, or even discussed?
01:59 idiosyncrat_ Well, in the SLIF DSL IIRC we allow both /* ... */ and per-line hash comments ("# ...").
02:00 idiosyncrat_ This seemed fine, I think.
02:12 rns idiosyncrat, ronsavage: re comment syntax for hiding --
02:13 rns to me, 'comment out' e.g. A ::= B (# C D #) means 'hide from syntax' (BNF in this case), while 'hide out' (e.g. A ::= B (- C D -) ) means 'hide from semantics' (evaluator in this case).
02:13 rns The former implies (and causes) the latter, but they are different.
02:17 Idiosyncrat joined #marpa
02:17 rns idiosyncrat: re http://irclog.perlgeek.de/m​arpa/2016-04-20#i_12363946 -- yes and that lookahead hackery looks very similar to randal.t -- comments are allowed inside pattern match (/ ... # / ) -- and the hopefully forthcoming AATM, does it not?
02:18 Idiosyncrat Well, the hiding syntax needs to be settled long before the AATM will be ready.
02:20 Idiosyncrat By the way, I just got a new Raspberry Pi 3, which I'm loading with Raspian right now.
02:21 rns re http://irclog.perlgeek.de/m​arpa/2016-04-20#i_12364267 -- sure, just noticed a similarity.
02:27 Idiosyncrat I'd tried a Raspberry Pi Model B Rev 2, with one 700 MHz core, and considered using it as a development machine, because I like to develop on a slow box.
02:28 Idiosyncrat ... it keeps me sensitive to performance issues.
02:28 Idiosyncrat But 1x700MHz was *too* slow.
02:29 Idiosyncrat The Pi 3 is 4x1GHz, so we'll see how that works out.
02:29 Idiosyncrat They say it's fast enough to run video decently.
03:21 rns left #marpa
05:48 ronsavage joined #marpa
07:22 koo7 joined #marpa
07:47 Idiosyncrat joined #marpa
10:44 kaare__ joined #marpa
12:34 harovali1 joined #marpa
13:44 pczarn joined #marpa
13:47 pczarn I suggest not adding both (- -) and (# #), because people won't build an intuition of which is which
13:50 pczarn instead of (# #), having /* */ style comments and some kind of one line comments is better
13:54 pczarn Further, there are other alternatives to (- -)
14:20 koo7 joined #marpa
15:19 kaare__ joined #marpa
16:40 kaare__ joined #marpa
17:06 idiosyncrat_ joined #marpa
17:07 idiosyncrat_ While I'm at it, I wonder if I shouldn't put Marpa::R3 under the MIT license.
17:08 idiosyncrat_ Does anybody remember the discussion the converting Libmarpa to the MIT license vs. Marpa::R2 which I kept LGPL'd.
17:08 idiosyncrat_ I'm thinking fresh start, I should be able to change the license.
17:13 idiosyncrat_ Also looking at https://github.com/jeffreykegler/kollos/blob/​master/components/main/kollos/grammar.lua.md --
17:14 idiosyncrat_ note the format, I think that would help folks go through the code -- it'd be nice to do something like that in the Perl, to aid readability.
17:18 idiosyncrat_ Re-reading https://groups.google.com/forum/#!msg/​marpa-parser/5IVTcnCqGwo/EfPbnbOOp6oJ
17:19 idiosyncrat_ My decision to keep Marpa::R2 LGPL was "for the time being", which now that Marpa::R2 is frozen, will probably be forever.
17:19 idiosyncrat_ But experience with the MIT license has been excellent, and the LGPL prevents some serious researchers from reading my code,
17:20 idiosyncrat_ so I think I will make the transition.
17:20 idiosyncrat_ Any comments?
17:24 idiosyncrat_ Or actually, shouldn't I put it under the Perl license?
17:31 idiosyncrat_ http://www.gnu.org/licenses/l​icense-list.html#PerlLicense
17:31 idiosyncrat_ Note this from the FSF: "We recommend you use this license for any Perl 4 or Perl 5 package you write, to promote coherence and uniformity in Perl programming."
19:14 kaare__ joined #marpa
19:43 kaare_ joined #marpa
20:29 kaare_ joined #marpa
20:50 harovali1 left #marpa
21:11 pczarn joined #marpa
22:40 ronsavage If I adopt a Perl package I keep its licence (sic, in Oz), but for my own I always use the Perl licence. But I was under the impression you would definitely move to MIT? Once I had a comment about my code/licence of the form "... unfortunately you've used the Perl licence" but frankly, I can't see what's wrong with it.
22:40 harovali joined #marpa
22:44 harovali hi !, Would some kind of grammar be good at finding patterns in a set of strings which compile under a primary grammar ? Or that enters in the non-cfg universe , and is always an ad-hoc special algorithm or attempt?
23:01 harovali I'd do something like this: have a semantics write down a log file with labels for every two nonterminals that happen to parse alongside, then sort the result and count repeated pairs. Alongside, rank nonterminals from most used to least used. Those pairs that occur most, which are composed of least used elements, will be the output of the pattern finding algorithm. Then look for a way to do it for three nonterminals, and for
23:06 harovali Then with the found pairs, or tuples, modify the original grammar to make some rules have special cases (or'd) , which would happen to be the patterns : )
23:39 harovali there is a lot of naïveness in my theory, I'd like to put it in the right words

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