Perl 6 - the future is here, just unevenly distributed

IRC log for #marpa, 2016-04-04

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

All times shown according to UTC.

Time Nick Message
00:06 koo7 idiosyncrat_, any plans in this direction?
00:06 idiosyncrat_ no, not really.
00:07 idiosyncrat_ you can pull them out of the progress reports already.
00:07 idiosyncrat_ you can also add nulling symbols and use nulling events
00:08 idiosyncrat_ That is, before every nonterminal whose prediction is of interest ...
00:08 idiosyncrat_ place a nulling symbol, and add a nulling event for it.
00:10 idiosyncrat_ And you should be able to use prediction events.
00:11 idiosyncrat_ Prediction events cannot be *marked* in the recognizer, only in the grammar, but ...
00:11 idiosyncrat_ they can be *activitated* and *deactivated* in the recognizer.
00:13 idiosyncrat_ (The reason for all this is there is some overhead per marked symbol for each event, so it pays to make the user declare which symbols he *might* want to use.  The user can then activate them and deactivate the events, depending on whether they actually want to see the events or not.)
00:14 idiosyncrat_ If you really can't narrow down the non-terminals for which you are interested in prediction events, you can mark all of them in the grammar.
00:28 ronsavage joined #marpa
00:45 koo7 thanks, will look into it
01:19 ronsavage 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
02:10 idiosyncrat_ I've pushed a new 'trial' version of Marpa::R3 to github
02:10 idiosyncrat_ All tests OK on my box.
02:10 idiosyncrat_ New in this version
02:11 idiosyncrat_ The "latm" adverb is now longer recognized by the SLIF.
02:11 idiosyncrat_ [name,values] is now the default action for rules.
02:12 idiosyncrat_ The POD has been modified accordingly -- though I may have missed some places where the defaults are referred to, so let me know.
02:14 idiosyncrat_ These 2 changes of the defaults, and yesterdays Unicode code were the "easy" fixes I wanted to get in first ...
02:14 teatime so latm is just the default now?
02:14 idiosyncrat_ and for once my predictions of what would be easy was right. :-)
02:15 idiosyncrat_ Not just the default -- there's no way to change it.
02:15 teatime kk
02:17 idiosyncrat_ Btw, all your current scripts will break on this new Marpa::R3, if they used the latm adverb ...
02:17 idiosyncrat_ which almost of them do.
02:19 idiosyncrat_ Just removing the "latm" adverb should get you going again.
02:20 idiosyncrat_ This one even caught me -- Marpa's SLIF is self-parsed -- to change the SLIF, I change a SLIF file ...
02:20 idiosyncrat_ and re-parse.
02:21 idiosyncrat_ This re-parse is not done every time a user parses a grammar, in part for effiiciency, but more to avoid the problems of circular development with breakage.
02:21 teatime heh.
02:22 idiosyncrat_ The SLIF self-description is parsed into an intermediate form, and that's what Marpa uses when you parse your own grammars.
02:23 idiosyncrat_ Anyway, I found that removing the "latm" adverb from the SLIF broke the SLIF's own grammar, which used the "latm" adverb ...
02:24 idiosyncrat_ and once again the intermediate form saved me. :-)
05:00 rns idiosyncrat: re http://irclog.perlgeek.de/marpa/2016-04-03#i_12280069 -- https://github.com/jeffreykegler/Marpa--R3/pull/3
05:01 idiosyncrat_ Yes, I'm reading it now
05:01 rns Ok, great.
05:02 idiosyncrat_ Now, explain for me ...
05:02 idiosyncrat_ the Config::Autoconf-generated configure *also* creates a Makefile.  Or doesn't it?
05:05 rns On windows, it seems like it doesn't and even if it did, there is no bash and friends to run it :(
05:06 idiosyncrat_ Your patch (as I read it) has me switch from the Config::Autoconf-generated Makefile to one generated by Makefile.PL
05:06 rns Hence the write_makefile.pl which writes Makefile.PL which creates Makefile for nmake
05:06 idiosyncrat_ But we don't just use the Config::Autoconf build for Windows.
05:07 rns I'm afraid we do:
05:07 rns my %perl_autoconf_os =     map { $_ => 1 } qw( MSWin32 openbsd solaris sunos midnightbsd );
05:08 rns Note MSWin32 -- or am I missing something?
05:08 idiosyncrat_ My point is I wonder how well this patch works for openbsd solaris sunos midnightbsd.
05:09 idiosyncrat_ That is, I wonder if it is assuming that Config::Autoconf builds are *just* for MSWin32, because at this point we use them for several other things.
05:10 rns Oh I see.
05:10 idiosyncrat_ jdurand IIRC is not only a Marpa build expert, but he also develops under Windows.
05:11 idiosyncrat_ I wonder what he thinks here.
05:11 rns Yes, Jean-Damien's input would be useful.
05:12 idiosyncrat_ Anyway, this is progress, but we'll have to look at this more carefully.
05:12 rns Sure, I tested it only under cygwin.
05:13 rns And windows, of course.
05:13 idiosyncrat_ Jean-Damien has suggested in the past that I convert to Config::Autoconf builds for all builds, and I didn't want to go that far, but I do want it to be possible to do a Config::Autoconf build on *every* target.
05:13 idiosyncrat_ rns: So you use *both* cygwin and MSWin32?
05:15 idiosyncrat_ jdurand: hope to hear from you on this, if you have the time.
05:15 rns Yes, cygwin (gcc 5.3.0, perl 5.22.1) under Windows XP (visual studio 2010 compiler, perl 5.20.1)
05:16 idiosyncrat_ Anyway, it's late California time.
05:16 idiosyncrat_ Good night!
05:17 rns Good night!
05:20 rns if somebody has access to openbsd and/or solaris and/or sunos and/or midnightbsd platform(s),
05:21 rns checking out https://github.com/rns/Marpa--R3/commits/fix_windows_build branch and running make releng to see if it's broken will be greatly appreciated.
05:43 rns left #marpa
06:10 jddurand joined #marpa
06:12 jddurand Config::Autoconf is generating config.h,  and is decoupled from Makefile*
07:39 ceridwen joined #marpa
09:02 koo7 joined #marpa
10:59 kaare_ joined #marpa
16:04 idiosyncrat_ joined #marpa
16:15 idiosyncrat_ rns: I thought over the Config::Autoconf thing and I realize I *had* this right in Marpa::R2.
16:15 idiosyncrat_ I'm going to redo it based on what's in Marpa::R2 and, once I'm done, I hope you'll retest.
16:44 rns joined #marpa
16:44 rns idiosyncrat: re http://irclog.perlgeek.de/marpa/2016-04-04#i_12284146 -- sure, just let me know,
16:45 rns . :)
16:53 idiosyncrat_ Btw, the reason that I wanted to do the conversion to EU::MM first thing in Marpa::R3, is so that we'd have the longest time possible to work out all the nits.
16:54 idiosyncrat_ I didn't want to wait until just before we planned a release to tackle all this stuff.
18:52 koo7 joined #marpa
19:07 ceridwen joined #marpa
19:30 kaare_ joined #marpa
23:16 idiosyncrat_ joined #marpa

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