Perl 6 - the future is here, just unevenly distributed

IRC log for #marpa, 2014-11-13

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

All times shown according to UTC.

Time Nick Message
00:09 jeffreykegler joined #marpa
00:15 lie_detector by chance, isnt https://jeffreykegler.github.io/Marpa-web-site/libmarpa_api/cpan_indexed/index.html available somewhere as a single html page?
00:21 jeffreykegler I think somebody did that -- it'd be in the IRC backlog
00:21 lie_detector k ill check
01:16 jeffreykegler lie_detector: enough folks have asked for single page HTML, that I'm looking into adding it to my build.
01:17 jeffreykegler It's something I'd occasionally use as well -- it is handy for searches
01:22 lie_detector would be nice
03:09 jeffreykegler The single page HTML for the Libmarpa API is now out there -- it's linked via my Marpa web page.
03:09 jeffreykegler ronsavage: this broke one of your links
03:11 jeffreykegler The one labeled "Libmarpa - The index for the docs" should now point to http://jeffreykegler.github.com/Marpa-web-site/libmarpa_api/stable/index.html
03:14 jeffreykegler Also flaviu's one-page HTML version of the Libmarpa, which you had under the label "Libmarpa - The docs on one page" can be replaced with
03:15 jeffreykegler http://jeffreykegler.github.com/Marpa-web-site/libmarpa_api/stable/api_one_page.html
03:15 jeffreykegler Thanks to ronsavage, lie_detector and flaviu.
03:21 lie_detector thanks!
04:12 ronsavage joined #marpa
04:22 ronsavage jeffreykegler: Link updated. Thanx!
12:17 lwa joined #marpa
14:09 lie_detector joined #marpa
16:06 LLamaRider joined #marpa
16:21 jeffreykegler joined #marpa
17:19 jeffreykegler daxim: I hope you say my reply yesterday re PEG, and that it was helpful.
17:39 jeffreykegler ronsavage: re Github issue #214 -- I've been holding off on closing pending your feedback, just in case my explanation missed the point.
17:57 jeffreykegler Immediate plan: fix Github issue #217 (a line/column numbering glitch found by Jean-Damien), and create a new developer's release of Marpa::R2 for CPAN.
17:58 jeffreykegler This release will catch the documentation up to the new features I added in the last release --
17:59 jeffreykegler rejection events, exhaustion events, $recce->last_completed_span() and the ability to change lexeme priorities on the fly.
18:20 daxim jeffreykegler, it wasn't helpful. I asked for test files, hoped to get a list of file names.  you answered "almost all of them contain left recursions", which is useless to me, because I do not have the knowledge to decide whether that is true for which test file.  I acked the t directory for "left" and "recurs", but that only gave uninteresting results.
18:25 daxim ronsavage, I took this photo of a marpa depiction at zhaoling, shenyang:  http://i.imgur.com/8vSETa9.jpg
19:47 jeffreykegler joined #marpa
19:51 jeffreykegler daxim: do you know who the other two figures are in the photo?
20:03 shadowpaste "jeffreykegler" at 217.168.150.38 pasted "Example of grammar PEG cannot handle" (23 lines) at http://fpaste.scsys.co.uk/440325
20:04 jeffreykegler I've pasted a very simple grammar that PEG cannot handle, due to left recursion.  It's taken from http://www.dalnefre.com/wp/2011/05/parsing-expression-grammars-part-4/
20:06 jeffreykegler The suggestion made for PEG users is to rewrite the grammar into a right recursion, then undo the right recursion in the semantic phase.
20:08 jeffreykegler This double rewrite is something which would have to be coded by hand.
20:13 jeffreykegler The topic (of what PEG cannot do) gets slippery because if you allow PEG to do post-parsing rewrites and adjustments.  If you do that PEG can do anything Marpa can do ...
20:14 jeffreykegler you use PEG to produce a series of tokens, and then run Marpa.
20:15 jeffreykegler I mean this seriously as thought problem because the post-processing can reasonably be seen as a piecemeal and ad hoc re-implementation of Marpa ...
20:15 jeffreykegler in which context it seems to me the reasonable thing to do is just use Marpa in the first place.
20:16 jeffreykegler I also mean it seriously in the sense that I have toyed with the idea of using PEG as an alternative lexer for Marpa -- PEG is powerful enough to be a reasonable lexer.
20:16 jeffreykegler daxim: was this more helpful?
20:21 jeffreykegler daxim: To answer the question more literally, every test file matching the pattern t/sl_*.t could not be duplicated using PEG, because the SLIF's DSL makes heavy use of left recursion, and is ambiguous as well.
20:24 jeffreykegler daxim: I'm sorry the answer turns out not to be simple, but the concept of "troublesome for PEG", is vague, and the PEG advocates move the boundaries around ...
20:25 jeffreykegler Moving the boundaries, I want to emphasize, is OK, *if* it results in more helpful techniques.  I'm not above shifting a boundary or not myself.
20:25 jeffreykegler But in this base, I believe the fork is pushing the food around the plate, and is not really good a picking it up.
20:26 jeffreykegler * in this base -> in the case
20:28 jeffreykegler * in this base -> in this case
20:41 flaviu joined #marpa
21:16 jeffreykegler One feature that could not be duplicated with PEG is Marpa's prioritized rules.  An example of one is in the SLIF DSL synopsis: https://metacpan.org/pod/Marpa::R2::Scanless
21:18 jeffreykegler These allow you to specify precedence, with both left and right association.  To do this, the rule has to be rewritten automatically and then parsed PEG could not do the rewrite-and-parse.
21:21 jeffreykegler If you're looking for rules of this kind, the double bar (||) in the DSL is the clue.
21:21 jeffreykegler Test files with prioritized rules of this kind are t/sl_astsyn.t t/sl_calc.t t/sl_external1.t t/sl_fullsyn.t t/sl_debug.t t/sl_dsl.t t/sl_ruby.t t/syn_stuifzand.t t/sl_ast.t t/dsl.t t/sl_gsyn.t
21:28 ronsavage jeffreykegler: re Github issue #214. Sorry. I'd like to see you comment on the issue in the docs, so readers would at least know a course of action is available.
22:27 ronsavage daxim: Thanx. I'm currently rebuilding my website to include that image......

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