Perl 6 - the future is here, just unevenly distributed

IRC log for #marpa, 2016-04-11

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

All times shown according to UTC.

Time Nick Message
01:28 JPGainsborough 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
03:08 ronsavage joined #marpa
03:28 idiosyncrat_ I've just uploaded Marpa-R3-1.001_000 to CPAN, mainly so I can see what cpantesters thinks of the new EU::MM-based build
03:31 idiosyncrat_ I went out of sequence -- there were Marpa-R3-3.003_00? releases uploaded earlier.  This confuses the cpantesters logic about last successfully PASS'ing release, which assumes the releases are in order.
03:32 idiosyncrat_ I may just bump the major release number, and make the next Marpa::R3 4.001_000
03:32 idiosyncrat_ I think the raw log of CPAN tests is available somewhere.  Does anybody remember where.
03:33 idiosyncrat_ It's useful for watching the first tests come in, to check if something went really wrong.
10:07 kaare_ joined #marpa
11:28 rns idiosyncrat: re http://irclog.perlgeek.de/marpa/2016-04-11#i_12316059 -- off the top of my head as it is, perhaps all tarballs except Marpa-R3-1.001_000 can be deleted and cpan indexer will do its job?
11:29 rns meaning 'Marpa::R3 4.001_000' just doesn't feel right somehow.
11:31 rns re the uploaded Marpa-R3-1.001_000 -- ok under cygwin, not ok under windows7 (PR filed), not ok under windows (PR filed), hopefully workable under thosemplatforms with the said patches.
11:40 rns jddurand: Jean-Damien, your input on https://github.com/jeffreykegler/Marpa--R3/pull/9 and https://github.com/jeffreykegler/Marpa--R3/pull/8 would be great to have.
13:37 rns idiosyncrat: re jirotka.t -- the progress so far -- code: https://github.com/rns/Marpa--R3/tree/sl_jirotka.t -- grammar converted to SLIF, it parses the input; call to the tested methods added, they print the result -- https://gist.github.com/rns/ed792656826264dd9530f44e56167931 -- except show_or_nodes() and show_and_nodes() which are in  Marpa::R3::Thin::R namespace and thus fail if called from as methods of Scanless::R instance
14:15 rns idiosyncrat: re timeflies.t and sl_timeflies.t -- the main difference IIRC is in how semantics is done.
14:17 rns sl_timeflies.t produces the result by traversing the ASTs returned by value() calls, while timeflies.t uses actions subs in package main.
14:21 rns If the timeflies.t way behavior is valuable as a test case then I can add it to sl_timeflies.t or, better, rename sl_timeflies to sl_timeflies_ast.t and convert timeflies.t to sl_timeflies.t (preserving action subs semantics).
14:21 rns This way they would be similar to sl_json.t and sl_json_ast.t as done currently.
15:02 idiosyncrat_ Actually these are all unindexed versions, so the CPAN indexer in not (yet) involved.
15:02 idiosyncrat_ rns: That was re http://irclog.perlgeek.de/marpa/2016-04-11#i_12317226
15:03 idiosyncrat_ The CPAN indexer refuses to index versions out of order.
15:04 idiosyncrat_ I think the thing for me to do here is to not fight the system, and bump the number major number to something above 3 -- probably 4.
15:08 rns Ok, I see.
15:09 idiosyncrat_ rns: re http://irclog.perlgeek.de/marpa/2016-04-11#i_12317779
15:10 idiosyncrat_ So you've got jirotka.t OK except for show_or_nodes() and show_and_nodes(), where I didn't complete the port the way I thought I had?
15:11 idiosyncrat_ Best to submit it as a pull request WITH THE FAILING TESTS and a comment reminding me of the situation ...
15:11 idiosyncrat_ then I can use it as a test case to make the fix.
15:12 idiosyncrat_ Even if the tests cause an abend, I think there is a Test::More workaround that allows the test to be in there anyway.
15:13 idiosyncrat_ OK, I found it -- it's 'todo_skip
15:15 idiosyncrat_ 'todo_skip' is documented here: http://search.cpan.org/~exodist/Test-Simple-1.302013_018/lib/Test/More.pm#Conditional_tests
15:16 idiosyncrat_ I'm about to run off for an appointment ...
15:17 idiosyncrat_ I will look at sl_timeflies.t, and the sl_catalan.t pull request later.  Thanks!
15:17 idiosyncrat_ AFK
16:10 rns idiosyncrat: re http://irclog.perlgeek.de/marpa/2016-04-11#i_12318135 -- https://github.com/jeffreykegler/Marpa--R3/pull/11
18:09 koo7 joined #marpa
18:22 rns As a safety precaution, all tests fail because, frankly speaking, I'm not quite sure what the expecte results should be.
20:11 idiosyncrat_ joined #marpa
20:17 idiosyncrat_ rns: re http://irclog.perlgeek.de/marpa/2016-04-11#i_12318805 -- Good thinking!
20:17 idiosyncrat_ Yes, it's safest if I look at the differences and pronounce the tests OK one by one, ...
20:18 idiosyncrat_ and that kind of caution is the whole point of having the tests in the first place.
20:18 idiosyncrat_ So, good catch.\
20:20 idiosyncrat_ In the last, the final '\' is a typo.  I'd probably better point that out, since anything random one types these days is probably some kind of emoticon.
20:26 idiosyncrat_ OK, I new have a CPAN testers matrix for the new Marpa::R3 -- http://matrix.cpantesters.org/?dist=Marpa-R3%201.001_000
20:27 idiosyncrat_ I'm tempted to say that it couldn't be worse, but that's not true -- some of the tests we specifically conducted before CPANtesters actually succeeded in CPANtesters, including all the appropriate Linux and Cygwin tests.
20:28 idiosyncrat_ But everything else failed, including Darwin which I specifically tested on my own Mac box, running Tiger.
20:29 * idiosyncrat_ keeps hearing from the Smithsonian which wants a running MacOs Tiger box for its collection, but he keeps putting them off
20:31 idiosyncrat_ I've got a lot of balls in my court, so if anyone wants to take a preliminary look to try to figure out any of these build issues, I'll be interested in what they find.  There are lots to choose from. :-)
20:33 idiosyncrat_ Note that Marpa::R3 now requires Perl 5.10.1 or greater.  I'd thought if I see MIN_PERL_VERSION in EU::MM, CPANtesters would not even attempt to test these, but apparently they give it a go anyway.  Does anyone know of a way to tell a CPANtester platform to just not bother?
20:35 idiosyncrat_ rns: On looking more closely I see that our failing MSWin result is yours: http://www.cpantesters.org/cpan/report/6cfd8343-647b-1036-b606-529979bada3d
20:35 idiosyncrat_ Was one or more of your recent pull requests to address that failure?
20:37 rns idiosyncrat: yes, this one https://github.com/jeffreykegler/Marpa--R3/pull/8 for windows xp and this https://github.com/jeffreykegler/Marpa--R3/pull/9 for windows 7 (sorry, didn't set up cpantesters there yet)
20:42 rns it's coming too late here, so have to go AFK, will backlog. Good night!
21:32 idiosyncrat_ rns: Thanks!
21:32 idiosyncrat_ A bit on the test suite for Marpa::R3, and my theory of the matter:
21:33 idiosyncrat_ Marpa::R3 (like Marpa::R2) makes extremely aggressive use of
21:33 idiosyncrat_ 1.) new mathematics;
21:33 idiosyncrat_ 2.) large amounts of bit-twiddling hand-optimized C code;
21:34 idiosyncrat_ 3.) Perl XS layer coding;
21:34 idiosyncrat_ 4.) Perl's build tools.
21:34 idiosyncrat_ In 1.) and 2.), I think Marpa must be among the 5 most complex modules on CPAN -- in 1.), it might be #1.
21:35 idiosyncrat_ With 3.) and 4.) there's more competition, but when I look for examples on CPAN to borrow ideas from, I have real trouble finding them.
21:36 idiosyncrat_ All of which is to lead to this point:
21:36 idiosyncrat_ in Marpa::R3, if it's a question of too many tests or too few, the users will usually thank us if we err on the side of too many tests.
21:38 idiosyncrat_ rns: I assume you'll backlog and will read this at some point.
21:40 idiosyncrat_ rns: re http://irclog.perlgeek.de/marpa/2016-04-11#i_12317974 -- yes, I think having both an sl_timeflies.t and a sl_timeflies_ast.t is a good idea.  If you'd like to proceed with that, I'd be grateful.
23:31 ceridwen joined #marpa

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