Perl 6 - the future is here, just unevenly distributed

IRC log for #marpa, 2016-04-15

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

All times shown according to UTC.

Time Nick Message
00:22 hobbs joined #marpa
01:08 rns joined #marpa
01:22 idiosyncrat_ re AATM: Obstacles
01:22 idiosyncrat_ 1.) What about lexeme events
01:22 idiosyncrat_ 2.) What about discards
01:23 idiosyncrat_ 3.) What about ranks
01:23 idiosyncrat_ For 1.) lexeme events, I think perhaps have post-lexeme events for all the lexemes -- that's easy.
01:24 idiosyncrat_ And pre-lexeme events -- which cause a lexeme to remain unread, cause *all* lexemes starting at that position to remain unread.
01:25 idiosyncrat_ For 3.) ranks are about eliminating it, and AATM about embracing it as a technique, so they don't really work well together, but ...
01:26 idiosyncrat_ perhaps just keep lexemes ranks as they are (they rank lexemes with the same start and the same end point) and allow mixing lexeme ranks and AATM for those who like to live dangerously.
01:27 idiosyncrat_ 2.) is really hard -- discards exclude certain spans of the input from being visible to the Earley sets; while overlapping lexemes in Libmarpa, which I'll need for AATM, require the input to be continuous.
01:28 idiosyncrat_ Initially, I may say that if AATM and ':discard' rules in the lexer are mutually exclusive -- one or the other, not both.
01:29 idiosyncrat_ If you use AATM (at least in the first cut) you'll have to account for whitespace in your grammar, perhaps using the interior/exterior strategy I implemented yesterday in t/sl_randal.t.
01:30 idiosyncrat_ This means AATM is an all-or-nothing choice in a grammar -- it applies to all lexemes or none.
01:31 idiosyncrat_ (This is unlike LATM used to be, where you could enable or disable it by lexeme.)
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
02:51 ilbot3 joined #marpa
02:51 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
03:40 ronsavage joined #marpa
03:59 idiosyncrat_ Down to 15 'REWORK' tests -- NAIF tests that need to be converted to SLIF before we can start the serious refactoring in Marpa::R3
04:00 idiosyncrat_ We started with 30, so this is the official halfway point. :-)
04:06 idiosyncrat_ Good night!
05:42 ilbot3 joined #marpa
05:42 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
07:21 ilbot3 joined #marpa
07:21 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
10:03 koo7 joined #marpa
10:59 rns joined #marpa
11:12 kaare__ joined #marpa
11:18 rns idiosyncrat: re AATM -- sounds good -- some liberty at a cost.
11:21 rns BTW, just thought randal.t looks behaves closely to AATM -- advance input one-by-one, scan by regex table, throw all scanned lexemes to marpa as alternatives, advance by the length of accepted token, repeat.
11:27 rns re REWORK -- just did ah2.t and bocage.t -> sl_ah2.t and chaf.t -> sl_gia.t per http://irclog.perlgeek.de/marpa/2016-04-14#i_12334613 -- can take more for the weekend.
12:01 kaare_ joined #marpa
12:08 kaare__ joined #marpa
12:32 ilbot3 joined #marpa
12:32 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
13:21 beaugunderson joined #marpa
14:17 ilbot3 joined #marpa
14:17 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
14:19 kaare_ joined #marpa
14:31 idiosyncrat_ joined #marpa
14:31 idiosyncrat_ rns: re http://irclog.perlgeek.de/marpa/2016-04-15#i_12342696
14:31 kaare__ joined #marpa
14:31 idiosyncrat_ Yes, creating a 2nd version of sl_randal.t will be a great initial test of the AATM
14:32 idiosyncrat_ rns: re http://irclog.perlgeek.de/marpa/2016-04-15#i_12342722 -- Thanks!
14:38 kaare_ joined #marpa
15:10 idiosyncrat_ rns: Incorporated your conversion of chaf.t to sl_chaf.t, which was perfect except for one thing, which was my fault ...
15:11 idiosyncrat_ I suggested folding it into sl_gia.t, but I was wrong about it fitting the paradigm for the sl_gia.t tests ...
15:11 idiosyncrat_ so I created an sl_chaf.t for it.
15:11 idiosyncrat_ In the future, if I suggest folding a test under a paradigm and it turns out to be a bad fit, just bounce it back at me, pointing out the mistake.
15:13 rns Ok, note taken.
15:17 maybekoo2 joined #marpa
15:20 user2_ joined #marpa
15:34 daxim_ joined #marpa
15:34 beaugunderson joined #marpa
15:41 rns idiosyncrat: re make -j4 issues ...
15:41 rns https://www.gnu.org/savannah-checkouts/gnu/autoconf/manual/autoconf-2.69/html_node/Parallel-Make.html#Parallel-Make
15:41 rns says: Support for parallel execution in make implementation varies. Generally, using GNU make is your best bet.
15:42 rns and then goes a length about what can be wring with *BSD make's
15:42 idiosyncrat_ Yes
15:44 idiosyncrat_ Does it say how to portably turn off parallelism?  I don't see anything about tha.
15:44 idiosyncrat_ s/about tha/about that/
15:46 rns thy are saying about -B:
15:46 rns You can avoid all these issues by using the -B option to enable compatibility semantics. However, that will effectively also disable all parallelism as that will cause prerequisites to be updated in the order they are listed in a rule.
15:56 idiosyncrat_ I don't mind disabling all parallelism
15:56 idiosyncrat_ I lived without quite happily for years :-)
15:58 idiosyncrat_ OK, there's a .POSIX target -- http://pubs.opengroup.org/onlinepubs/009695399/utilities/make.html
15:59 idiosyncrat_ If we specify that, it should turn off parallelism.
16:00 idiosyncrat_ Because I wonder if -B doesn't just shift the problem elsewhere -- I don't see that the -B option is portable.
16:00 idiosyncrat_ Could you test the .POSIX target.  I hope it's compatible with EU::MM generated makes.
16:05 idiosyncrat_ I also note we can rely on $Config{MAKE} telling us if we are running dmake or nmake -- http://search.cpan.org/~bingos/ExtUtils-MakeMaker-7.10/lib/ExtUtils/MakeMaker.pm#MAKE
16:05 idiosyncrat_ I think it's fine to special case for those, relying on $Config{MAKE}
16:19 rns re .POSIX -- Include the special target .POSIX  Omit any special target reserved for implementations (a leading period followed by uppercase letters) that has not been specified by this section
16:20 rns http://pubs.opengroup.org/onlinepubs/009695399/utilities/make.html
16:21 rns which leaves us with .DEFAULT .IGNORE .POSIX .PRECIOUS .SCCS_GET .SILENT .SUFFIXES
16:21 rns I see we use .PHONY
16:21 rns Sorry, have to go , will look deeper into it this weekend. WIll backlog
16:21 rns AFK
16:21 rns left #marpa
16:23 idiosyncrat_ Yes, since we have to use the Makefile's generated by EU::MM, we may not be able to rely on no special targets being generated, in which case .POSIX will not help us.
16:36 Pursuit joined #marpa
17:12 VsyachePuz joined #marpa
17:54 ilbot3 joined #marpa
17:54 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
18:03 idiosyncrat_ Re AATM:
18:03 idiosyncrat_ On the way to implementing this, I'd need to change the way the SLIF does input.
18:04 idiosyncrat_ Right now, the input must be read all-at-once, at the beginning, and cannot be changed afterwards, but ...
18:04 idiosyncrat_ you can move around in the input string.
18:05 idiosyncrat_ This is a somewhat idiosyncratic approach, which was easy to implement at the time, and so which is in Marpa::R2 for historical reasons.
18:05 idiosyncrat_ For AATM, allowing jumping around in the input will be too complex, but ...
18:06 idiosyncrat_ I can change things so that the input need not be read all at once -- so that you can extend it.
18:06 idiosyncrat_ Allowing you to extend the input will help some current users -- in some applications it can be inconvenient to have to specify your entire input at the beginning.
18:07 idiosyncrat_ But I will now prohibit moving back and forth in the input string, which might be seen as a regression.
19:27 Pursuit joined #marpa
19:43 idiosyncrat_ AFK
22:35 ronsavage joined #marpa
23:31 idiosyncrat_ joined #marpa

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