Perl 6 - the future is here, just unevenly distributed

IRC log for #marpa, 2014-05-03

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

All times shown according to UTC.

Time Nick Message
00:04 jeffreykegler joined #marpa
00:05 jeffreykegler LLamaRider: picking up on your having to "deal with" autoconf in order to build libmarpa ...
00:06 jeffreykegler There's two senses of this -- one is to have to deal with it in the sense of setting up the configuration files, and ...
00:06 jeffreykegler the second is deal with it in terms of running configure, with standard arguments, followed by make with standard targets.
00:07 jeffreykegler You'd have to deal with autoconf in the 2nd sense, which is supposed to be pretty easy -- after that's why we put up with the misery of creating autoconf configurations, because of the promise they'll port all over the place and "just run".
00:08 jeffreykegler So if you want to build a libmarpa shared library and the configuration file doesn't "just run", it's a Libmarpa bug which you can bounce back at me.
00:13 ronsavage joined #marpa
00:23 ronsavage Ah, I mis-wrote. It's CFL's - Context Free Languages.
00:47 jeffreykegler joined #marpa
00:48 jeffreykegler ronsavage: yes, did you ask a question re CFL's?  It's not on the backlog
01:13 ronsavage Yep. Quoting myself: "(18:44:44) ronsavage: And CLF's are?".
01:14 ronsavage Also possibly lost was my agreement to amend the docs. The FAQ now is: http://savage.net.au/Perl-modules​/html/marpa.papers/chapter3.html
01:16 jeffreykegler ronsavage: seems like the log-bot is snubbing you
01:17 jeffreykegler CFL == context-free-language.  A CFL is something you can write in BNF and vice versa.
01:17 jeffreykegler Hence the connection to Marpa which is a general BNF parser -- that is, it parses anything you can write in BNF and therefore ...
01:17 jeffreykegler any CFL.
01:18 jeffreykegler Which makes Marpa very different in scope from recursive descent, yacc/bison, regular expressions, etc., etc.
01:28 shadowpaste "jeffreykegler" at 99.118.55.44 pasted "For FAQ: What is Libmarpa?" (6 lines) at http://scsys.co.uk:8002/363639
01:28 jeffreykegler ronsavage: I've drafted an expanded version of the FAQ, which I just pasted.
01:29 jeffreykegler I was a bit lazy when it came to the links, but I assume that's an easy cut-n-paste on your side.
01:30 jeffreykegler Also, maybe it should go to the end since it is (at this point at least) of interest mainly to advanced users.
01:49 ronsavage Thanx. I've updated the FAQ. I left the libmarpa note at the start because it's short and can be skipped easily by beginners
01:50 jeffreykegler Great!  Thanks!
01:51 ronsavage Also, doing that suits the way I think: Building blocks first, then superstructure.
01:51 ronsavage Off to lunch!
06:49 ronsavage joined #marpa
08:06 ronsavage joined #marpa
08:06 yxhuvud joined #marpa
08:06 shadowpaste joined #marpa
08:06 sivoais joined #marpa
08:06 lucs joined #marpa
15:03 jdurand joined #marpa
15:34 jdurand Jeffrey, looking how marpa_version() is implemented, maybe you could state in the doc that this method always succeed (?)
16:51 jeffreykegler joined #marpa
16:54 jeffreykegler jdurand: re marpa_version().  If I state it always succeeds, but then some day find I need to detect an error and return it, I could not do so without breaking backward compatibility.
16:55 jeffreykegler The very few states I've documented a method as always succeeding, I've come to regret it.
16:55 jeffreykegler * "very few states" -> "very few times"
17:30 jeffreykegler Marpa-R2 2.085_006 is up on CPAN.  It has a lot of rearrangements and repackagings, centered on shared library support.  Testing is appreciated!
18:07 jdurand joined #marpa
20:29 Guest63808 joined #marpa
20:33 vivtek I noticed on today's log that "dealing with" autoconf was an issue -
20:34 vivtek it bit me as well. Config::Autoconf required force install (as did
20:34 vivtek Marpa itself.
20:36 vivtek http://www.cpantesters.org/cpan/report​/cf9fa0d8-6c02-1014-af2c-da6ebb10c5d0
20:36 vivtek Really odd, as I just don't see anything in t/sl_advent.t that could possibly have caused those errors.
20:36 vivtek Ya gotta love Perl on Windows, though.
20:43 jdurand What was the problem with Config::AutoConf ?
20:56 vivtek Hmm, let me check.
20:58 vivtek http://www.cpantesters.org/cpan/report​/2029882c-6c02-1014-bb54-4681bb10c5d0
20:58 vivtek (One nice thing about sparse testing on Windows for CPANtesters is it's relatively easy to find your own test reports.)
20:58 vivtek It identified three utilities correctly, but for some reason thought they weren't executable. I didn't investigate further,
20:59 vivtek beyond confirming that one or two *were* executable. I wanted to play with Marpa, so I just assumed everything was kosher enough.
20:59 vivtek After all, the utilities *had* been found.
21:00 jdurand Remove temporarlw cygwin from your path, it should install ok then
21:01 vivtek Ha. It installed fine with force. Much simpler than temporarily messing with a perfectly functional path.
21:01 jdurand No pb.
21:01 jdurand I personnally destroyed Cygwin on my PC: this is causing too much trouble
21:02 jdurand In particular with all programs that think that there is a shell available, and when they test it like autoconf, the shell tests are failing because there is no ral tty behind
21:02 vivtek Yeah, it get get funky, I agree. I have an eclectic mishmash of Cygwin and Ming32 on a 64-bit Windows. Occasionally things work fine! :-)
21:02 jdurand Ok. I personnaly moved 100% to GnuWin32 packages.
21:03 jdurand Everything is there and it works perfectly
21:03 vivtek I'll have to check it out sooner or later.
21:03 vivtek I usually get excited about this kind of thing about every three or four years, and then everything is completely different from the previous iteration.
21:03 jdurand Yep, recommended, truely. This is impressive.
21:04 jdurand Anyway, you got Marpa on your PC that's good -;
21:04 vivtek Yeah, I even wrote a tool for it already!
21:04 vivtek MarpaX::Tester - I was driving myself crazy.
21:05 jdurand I went to https://metacpan.org/pod/MarpaX::Tester
21:05 jdurand That's a very good idea, and seems well done from the POD point of view
21:05 jdurand Not tested yet though
21:05 vivtek Well, I just uploaded it about three hours ago.
21:06 vivtek The documentation could use a little more fleshing out, but I'll get to that soon.
21:06 vivtek It sure helped me find one bug already - if you can iterate through a bunch of different tries in a couple of minutes it gets a *lot* easier to find stuff.
21:07 jdurand No pb - that's really very very new then! Looking to the source, "use 5.006;" is useless. Marpa already forces a "use 5.010"; as far as I remember
21:07 vivtek Boilerplate.
21:08 jdurand Re tests - I agree, this is a very nice idea, truely
21:08 jdurand welcome to Marpa world btw -;
21:08 vivtek I've been meaning to take the plunge for a couple of years - since I first read about it, actually.
21:08 vivtek I just finally convinced myself this week was the week. :-)
21:11 vivtek Looks like I screwed up the manifest, though - didn't put the command-line script into it. Stupid.
21:11 vivtek Ah well. That's why we have CPANtesters.
21:11 jdurand No pb! It is never too late. Marpa popularity is increasing slowly but surely. Its perl-only interface on one side, very low-level API on the otherside, is leaving place for other interfaces.
21:12 jdurand I am on the my personnal second version of another C wrapper to it -;
21:12 vivtek Oooooh, Marpa and me are going to be very, very good friends indeed. I have lots and lots of NLP tasks I want to try with it, and I think it should work - it's designed to deal well with ambiguous parses.
21:14 jdurand Good, then... looking forward to your work with it - do not forget to publicize if you feel it deserves it - blogs.perl.org in case of Perl work -;
21:15 vivtek My new year's resolution has been to write more about code. So far I've published a couple of fairly meaty articles to my site and two posts to blogs.perl.org.  I've got one on the front page there tonight, actually, on working with KeePass.
21:17 jdurand Yep, I see on https://metacpan.org/author/MICHAEL you have several modules indeed, very good and have a nice work with us! Jeffrey answers are usually the bests but other try to cope with his level -;
21:31 vivtek Whew. OK. v0.02 should work better. What a boneheaded mistake! I really have to do a better boilerplate handler for this kind of stuff; I'm bad with detail.
22:20 jdurand Jeffrey, upon the events are http://jeffreykegler.github.io/Marpa-web-site/l​ibmarpa_api/latest/Event-codes.html#Event-codes, is MARPA_EVENT_NULLING_TERMINAL a fatal event or a warning event ? Grammar.pm seems to consider this is fatal.
22:20 jdurand "upon the events at"
22:33 jeffreykegler joined #marpa
22:34 ronsavage vivtek: In your module Tree::Walker it refers to Data::Table::Lazy, but I could not find any such module on MetaCPAN. I'm confused :-)
22:35 jeffreykegler jdurand: re http://irclog.perlgeek.de/​marpa/2014-05-03#i_8674665 -- the Libmarpa API description of marpa_g_precompute() gives the full details on this ...
22:35 jeffreykegler Nulling terminals are detected at grammar precomputation time -- they are always fatal errors.
22:36 vivtek Yeah, I need to get that polished up at some point, Ron. Baby steps.
22:36 jeffreykegler The events mechanism is to allow the upper layer to get a full list of them to report to the user, so the user does not have to fix them one at a time.
22:37 jdurand Jeffrey, yes you're right. Sorry.
22:38 jeffreykegler No problem.  I don't expect anyone to remember every corner of the Marpa documentation.  I certainly don't.
22:40 ronsavage Marpa::R2 V 2.085006
22:40 ronsavage Counts: Tests: 542. Modules: 8. Passes: 8. Fails: 0
22:40 ronsavage Duration: 1 minute and 35 seconds
22:40 jeffreykegler ronsavage: Thanks
22:42 ronsavage vivtek: No problem. I was interested because I took over maintenance of Tree::DAG_Node, which has walk_down().
22:43 ronsavage Uh-oh. Auto-emoticons. I mean Tree :: DAG_Node.
22:43 vivtek Oh, I think I remember that one. Cool! Always nice when stuff gets taken over when it needs it.
22:44 ronsavage jeffreykegler: I did test and post the previous version but did not see your response. I assume you saw mine and your reply got gobbled up by the log-bot
22:44 jeffreykegler vivtek: Welcome
22:44 vivtek Thanks! I've been watching your work for the past couple of years. Bang-up job!
22:45 jeffreykegler vivtek: Thanks
22:45 ronsavage vivtek: Tree :: DAG_Node is the sort of module you might mention (hint, hint) in your SEE ALSO
22:48 ronsavage vivtek: I've just added Tree::Walker to the SEE ALSO of Tree :: DAG_Node
22:49 vivtek Change made and pushed to Github - no promises when I'll be updating Tree::Walker, though - I've been using it for site publishing (to walk component files) but it's been pretty stable lately. Eventually those chords will come back around, though. They always do.
22:51 vivtek Oh, but now I remember Tree::Walker is failing some CPAN tests due to an omission in the POD. I'll have to push a new version soon.
22:56 vivtek G'night, all. Past midnight here and I'm going to regret that come morning....
22:58 jdurand Bonne nuit vivtek -;

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