Perl 6 - the future is here, just unevenly distributed

IRC log for #marpa, 2016-07-11

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

All times shown according to UTC.

Time Nick Message
03:02 ronsavage JK: In metag.bnf there are 2 forms for start rule. Will the one not using ':start' be deimplemented?
03:26 idiosyncrat_ I plan to keep them both.
03:26 idiosyncrat_ Taking the lexically first rule as the start rule is cleaner, and is what I expect will almost always be done.
03:26 idiosyncrat_ But an explicit ':start' can be handy for automatic generation of grammars.
03:43 ronsavage JK: Thanx, but what I meant was, given form 1 "<start rule> ::= (':start' <op declare bnf>) symbol", form 2 "<start rule> ::= ('start' 'symbol' 'is') symbol", will you drop form 2?
04:30 idiosyncrat_ Don't know
04:31 idiosyncrat_ Actually, I like form 2 better than form 1.
04:31 idiosyncrat_ There's a logic behind form 1, but I don't find it particularly mnemonic.
04:32 idiosyncrat_ And since that logic is my invention, if I don't find it memorable, I very much wonder if anyone else does.
04:34 idiosyncrat_ === On another topic ===
04:34 idiosyncrat_ I've added some recursive logic to the Perl XS code.
04:35 idiosyncrat_ This is the first time in the almost 9 year history of Marpa I've used recursion which, when you consider the problem Marpa addresses is not a small statement.
04:36 idiosyncrat_ Always before if I needed the equivalent of recursion, I created a special stack so that I did not recurse on the call stack.
04:37 idiosyncrat_ My reason was to allow Marpa to be run in certain threaded environment which had extremely small upper limits for call stack size.
04:38 idiosyncrat_ When I started Marpa I had no idea who my adopters would be -- for all I knew 90% of its usage might be in threaded environments with tiny stacks. :-)
04:40 idiosyncrat_ At point I think it's safe to allow Marpa to make reasonable, careful use of recursion on the call stack and
04:40 idiosyncrat_ it's not reasonable to do a lot of special programming to make it fit into unusual environments.
04:41 idiosyncrat_ The application, by the way, is the conversion of Lua tables to Perl hashes.
04:41 idiosyncrat_ Lua tables may contain other Lua tables, and it's one of many problems in the Marpa area which lend itself to recursive calls as a solution.
04:42 idiosyncrat_ Good night!
05:57 sivoais joined #marpa
06:34 ronsavage JK: Re http://irclog.perlgeek.de/m​arpa/2016-07-11#i_12818285. I'm the opposite. For me, form 1 is much clearer.
06:35 ronsavage Form 2, OTOH, smacks of Cobol!
07:21 sivoais joined #marpa
07:34 sivoais joined #marpa
09:10 maybekoo2 joined #marpa
10:47 kaare_ joined #marpa
13:09 rgrinberg joined #marpa
15:56 sivoais joined #marpa
16:09 sivoais joined #marpa
17:05 VsyachePuz joined #marpa
17:32 rgrinberg joined #marpa
18:10 sivoais joined #marpa
18:29 pczarn joined #marpa
18:30 pczarn I agree, form 1 is clearer.
18:39 rgrinberg joined #marpa
20:34 aredridel joined #marpa
21:05 idiosyncrat_ joined #marpa
22:46 ronsavage joined #marpa
23:49 ronsavage joined #marpa

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