Perl 6 - the future is here, just unevenly distributed

IRC log for #marpa, 2014-08-26

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

All times shown according to UTC.

Time Nick Message
00:14 ronsavage1 The final re-write, hopefully, of my new article: http://savage.net.au/Ron/html/A.New.Marpa-based.Parser.for.GraphViz.html
00:21 jeffreykegler ronsavage1: Is the code somewhere online?
00:22 jeffreykegler I notice the events for close and open braces/brackets.  Is that for special lexical processing of some sort?
00:29 ronsavage1 Yes. {} flag a new subgraph, and I output them and indent the tree. [] flag attributes for classes of nodes, edges and (sub)graphs, and individually for node and edges. I don't output them - the [] - , but I do indent the tree. And since attribute values can have strings with embedded spaces, I manually handle that. Hence the whole adoption of event-style processing.
00:31 jeffreykegler So you got all that working nicely?  I don't think there are that many examples of Marpa-event-based processing out there, and this is probably the best documented at this point.
00:31 ronsavage1 Thanx.
00:32 ronsavage1 I've just re-designing part of my web site. It will, funnily enough, more strongly emphasize Marpa, whose bandwagon I'm definitely on.
00:33 jeffreykegler An alternative would be to generate the Marpa AST, and post-process it, though that would not necessary be the best way to do the embedded spaces.
00:35 ronsavage1 OK. And as for other event articles, there was http://savage.net.au/Ron/html/Conditional.preservation.of.whitespace.html, in which I've discovered a subtle bug or two, but that sort of thing is inescapable.
00:55 jeffreykegler ronsavage1: re your website: would you like to become the "unofficial official" Marpa website?
00:55 jeffreykegler "Unofficial" because you retain editorial control.
00:57 jeffreykegler A complaint Gabor make publicly recently is that we are perhaps the most prominent Perl project without a website --
00:58 jeffreykegler which is not strictly true, because there is mine, but I sort of understand what Gabor was saying --
01:00 jeffreykegler my own website, by its nature, is also going to be heavily oriented toward the technical, the mathematical, and the bleeding edge, in a way that does not make it the 100% best public face for Marpa.
01:07 jeffreykegler *also -> *always
01:07 ronsavage1 Sure - I can do that. You mean this one (https://jeffreykegler.github.io/Marpa-web-site/) right?
01:08 jeffreykegler That's remain mine, is my idea, and in the future we'll direct folks to your site as a first stop, with mine for those who have decided to do more exploring.
01:09 jeffreykegler So that means it would be up to you to come up with a new public Marpa landing page -- which could be a page you already have out there.
01:10 jeffreykegler To make a comparison, it's kind of the "user'
01:10 jeffreykegler s manual" vs. specification type thing --
01:10 jeffreykegler where your page is sort of like the user's manual, while mine is more like the specification.
01:13 ronsavage1 OK. And right now it is appropriate for me to activate such a page. Errr, when I fix up a db which is used to generate one of my web pages.
01:14 ronsavage1 Do you see this page a having a set of links such as The Marpa Papers, etc?
01:14 jeffreykegler Yes
01:15 jeffreykegler A good first page for someone new to Marpa,
01:15 jeffreykegler which is I assume what you were about to do any way.
01:16 jeffreykegler Because, face it, my "official official" page plunges the reader very quickly into the details, a bit too quickly for a 1st landing page.
01:29 jeffreykegler ronsavage1: Of course, I'm assuming that in some visible position, you'll have the link to "Jeffrey's official Marpa page", for those who are looking for that kind of thing.
01:35 ronsavage1 Sure. Chap 1 of the MPs has a list of links. I'd duplicate some of them on the landing page.
01:38 jeffreykegler lucs: when ronsavage has the page ready, could it go in the IRC channel's topic?
01:39 jeffreykegler That's getting kind of crowded, but perhaps the "Source code: " entry could be dropped, on the idea that given Ron's page and mine, one should be able to get to the Github repository in two clicks.
01:43 lucs jeffreykegler: Sure, good idea.
03:54 ronsavage joined #marpa
03:54 ronsavage OK. The landing page is at http://savage.net.au/Marpa-and-Graphviz.html
04:18 jdurand joined #marpa
04:20 jdurand Re http://irclog.perlgeek.de/marpa/2014-08-25#i_9242382 - the other way, less intrusive, is to put everything togeher in a .bat file, in this file you will put all your environmenet setup. Put this file on your desktop, then just click on it -;
05:01 rns joined #marpa
05:09 jeffreykegler joined #marpa
05:10 jeffreykegler ronsavage: :Looks good, but could you separate the Marpa and the Graphviz pages?
05:11 jeffreykegler As it stands, a beginner would think Graphviz and Marpa are very closely tied together.
05:12 jeffreykegler It may leave (for now) a very small Marpa page, but more can be added to it.
05:17 rns jeffreykegler: re http://irclog.perlgeek.de/marpa/2014-08-25#i_9240634 and http://irclog.perlgeek.de/marpa/2014-08-25#i_9240650 — all true
05:18 rns I was just building from an obsolete repo. Now I see that building Marpa::R2 doesn't require cweb and tex so what I've asked for had already been done, thanks.
05:28 jeffreykegler rns: Great!
05:32 jeffreykegler ronsavage: In the Marpa page, I'd like a list of tutorials -- I can gather this if you'd like.  It'd be those I mention at the end of my blog posts, plus the tutorial section in my annotated blog index.
05:36 jeffreykegler I'd also like to start crowdsourcing a list of Marpa applications, those which are beyond the experimental or prototype stage, and which have real users -- I think there's a few at this point -- your work being the first of these.
05:36 jeffreykegler For example, I know that IBM has gone public about using Marpa, and can find the link.
05:37 jeffreykegler And there was Ruslan Zakirov's recent email to the G+ group.
05:40 jeffreykegler ronsavage: Here's the IBM page: https://www.ibm.com/developerworks/community/blogs/jalvord/entry/sitworld_itm_situation_audit?lang=en
05:43 sivoais has anyone thought about combining Marpa with evolutionary algorithms? I've been mulling over that for a while, but I haven't had time to play with it.
05:43 jeffreykegler I know of other companies using Marpa, via bug reports, but I want to be discrete and not announce that people and/or organizations are Marpa users who don't want that kind of notice, even if they've filed a bug report or asked a question in a forum that's open to the public.
05:44 jeffreykegler sivoais: what's an evolutionary algorithm?
05:45 sivoais jeffreykegler: for example, genetic algorithms from AI and machine learning
05:45 jeffreykegler I found the Wikipedia page.
05:46 jeffreykegler That is certainly an interesting question, and my answer as to whether Marpa can help is .....
05:46 jeffreykegler I haven't a clue.
05:47 jeffreykegler On one hand, what is done with DNA is a kind of parsing, and Marpa should be able to assist with that.
05:47 jeffreykegler On the other hand, it's a very specialized type, and whether Marpa adds real value is not immediately obvious to me.
05:48 sivoais here's something closer to what I'm thinking about <http://en.wikipedia.org/wiki/Grammar_induction>
05:49 jeffreykegler Hmmmmm
05:50 jeffreykegler I'd want to think about this, and perhaps glance at the current work in the field.
05:50 sivoais I don't have the tuits to play with the idea right now, but I'd like to see it done. Practical uses include automatically extracting out structured data from, say, search results on Amazon.
05:51 sivoais someone pointed me to <https://import.io/> and that prompted me to think about it again
05:52 rns http://www.sequitur.info/ (mentioned in the above wikipedia page) is a working algorythm for grammar induction.
05:53 rns it treats an n-gram (n>=1) which occurs more than once as a non-terminal and builds a representation which can be serialized to a bnf grammar.
05:54 sivoais mmm, interesting
05:54 rns this is provided by https://code.google.com/p/sequitur/ (Full C++) implementation
05:55 rns I once experimented with a Perl implementation — https://github.com/rns/Algorithm-Sequitur
05:57 jeffreykegler Sequitur looks a lot like CYK parsing,
05:57 jeffreykegler which seems to be an observation that's been made in the literature.
05:58 jeffreykegler In AI, I think it's possible Marpa may be able to go head-to-head with CYK parsing, losing some battles and winning others.
05:58 jeffreykegler I think this one may go to CYK parsing.
05:59 jeffreykegler CYK parsing essentially "brute forces" the structure from the bottom up.
06:00 jeffreykegler Which in cases where you don't have a lot of guaranteed overall structure to help you prune the possibilities, is a winning strategy.
06:00 jeffreykegler Marpa uses top-down pruning to help out its bottom up portion.
06:02 jeffreykegler In this case, however, there doesn't seem to be much you can do in the way of top-down pruning.
06:03 jeffreykegler sivoais: it's an interesting thought, thanks for bringing it up.
06:04 jeffreykegler It's 11:04 PM California time, so AFK
06:05 rns jeffreykegler: good night and good luck. :)
06:06 sivoais late here too, night o/
06:11 rns jeffreykegler: re http://irclog.perlgeek.de/marpa/2014-08-25#i_9242382 — here is what MS does for their visualc env vars:
06:13 rns have a bat file with all the settings (you can put there the commands specified by JD) and then have a 'Strawberry Perl shell' shortcut on your desktop or in the menu
06:18 rns In Target: field, specify %comspec% /k "c:\path-to-your-settings-bat-file\settings.bat" and this will open a command prompt with all env vars set as per the settings bat file.
06:21 rns Another (less GUI) way is to create a settings.bat file with all commands setting env vars and then a run.bat with the following contents:
06:23 rns setting.bat && %*
06:24 rns Then if you type run <command>, e.g. run perl -v on the command line, 'perl -v' will see all the env vars set by settings.bat.
06:25 rns Overall, it's a good idea to create a dir like c:\bat and put it in the PATH by using the GUI and then do everything else related to env settings by creating bat files in that dir.
06:25 rns Hope this helps.
07:49 ronsavage joined #marpa
07:56 ronsavage I'll split the Marpa and Graphviz pages. It'll be easy.
08:03 rns joined #marpa
08:04 rns jdurand: re http://irclog.perlgeek.de/marpa/2014-08-23#i_9232783 :)) easy — I just don't use cygwin for anything except Marpa. :)
10:09 ronsavage OK. A standalone page for Marpa: http://savage.net.au/Marpa.html
10:13 ronsavage1 joined #marpa
10:19 rns joined #marpa
10:20 rns joined #marpa
10:23 rns test from adiIRC, please ignore.
10:29 rns joined #marpa
15:50 rns joined #marpa
15:51 rns ronsavage: re http://irclog.perlgeek.de/marpa/2014-08-26#i_9243938 — looks very good: clear, consice and no-nonsense.
15:52 rns s/consice/concise/
16:46 Topic for #marpa is now Start here: http://savage.net.au/Marpa.html - Pastebin: http://scsys.co.uk:8002/marpa - IRC log: http://irclog.perlgeek.de/marpa/today
16:47 lucs ronsavage1: Maybe a link to the source code would be nice on your page? (https://github.com/jeffreykegler/Marpa--R2)
17:21 jeffreykegler joined #marpa
18:32 jeffreykegler joined #marpa
18:36 shadowpaste "jeffreykegler" at 162.232.214.245 pasted "Marpa description for web site" (28 lines) at http://scsys.co.uk:8002/418882
18:36 jeffreykegler I've just pasted my suggestion for a Marpa description at the top of the new Web page.  Comments welcome.
18:37 * jeffreykegler has a flaky Internet connection at the moment
18:38 jeffreykegler lucs: re the topic.  Thanks.  Could you add my web site back in, calling it "Jeffrey's Marpa site"?
18:46 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
18:46 lucs There we go.
18:49 jeffreykegler joined #marpa
18:59 jeffreykegler joined #marpa
21:57 ronsavage joined #marpa
22:00 ronsavage1 joined #marpa
22:04 ronsavage joined #marpa
22:12 ronsavage1 joined #marpa
22:51 ilbot3 joined #marpa
22:51 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
22:57 ronsavage joined #marpa
22:58 ronsavage I've added the url of the github repo.
23:02 ronsavage1 joined #marpa
23:50 jeffreykegler joined #marpa
23:54 jeffreykegler ron
23:55 jeffreykegler ron: what's a good format for suggesting additions to your web page?  Copy the HTML and hack it?  Or is it on Github?
23:58 ronsavage1 It's stored in XML and built by a Perl script. Probably best just to email me text.
23:58 ronsavage1 Actually, I had never thought of putting it on github. I'll do that now.
23:59 ronsavage1 Also, pidgin says you're last msg was at 23:58. Is that you're local time? I'm typing that at 09:59.

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