Perl 6 - the future is here, just unevenly distributed

IRC log for #marpa, 2016-04-05

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

All times shown according to UTC.

Time Nick Message
03:41 idiosyncrat_ re the builds
03:42 idiosyncrat_ Some trouble is caused by the fact that builds of libmarpa must be position-independent-code (PIC), but static libraries.
03:44 idiosyncrat_ Static because that's what is required for a component that needs to be part of another library.
03:44 idiosyncrat_ PIC because the final destination is a shared library.
03:46 idiosyncrat_ This is an unusual requirement, since static linking does not require PIC, but it does in this case because the final destination is a shared library.
03:46 ceridwen joined #marpa
04:36 ronsavage joined #marpa
04:37 ronsavage I'm finally on my old laptop, after the wireless chip on the new laptop died.
04:56 idiosyncrat_ A new 'trial' branch of Marpa::R3, with the latest reworking of the Config::Autoconf build logic.
04:57 idiosyncrat_ ronsavage: Good luck
04:57 idiosyncrat_ and good night!
04:58 rns idiosyncrat: great, I'll take a look.
05:37 ronsavage left #marpa
05:40 rns idiosyncrat: re http://irclog.perlgeek.de/m​arpa/2016-04-05#i_12287087 -- builds ok under windows for me.
05:43 rns Except that nmake test, once it ran all tests in t/*.t, cd's to xs dir and runs nmake test again and, as test section is SKIP'ped in xs/Makefile.PL, produces ‘U1073: don't know how to make 'test'’
05:44 rns The recommended workaround is specifying test as a rule (description block in nmake-speak) -- https://msdn.microsoft.com/​en-us/library/65faht2w.aspx
05:45 rns A possible fix is in this commit https://github.com/rns/Marpa--R3/commit/​f8f38c8526b9162e6f2fdba17f0d446376c62394
08:31 koo7 joined #marpa
11:31 kaare_ joined #marpa
13:02 koo7 ok, so, i have it print out the next expected symbol, i think
13:02 koo7 (the grammar is rather messy)
13:02 koo7 the items of the progress report
13:04 koo7 and im thinking, this is useful, but what i also want is to know the possible evaluation trees, given that the input was finished
13:04 koo7 everything has to parse to the start symbol
13:05 koo7 but if my input so far is "if true th", i want to know that this can parse into an if statement
13:24 sadmac joined #marpa
13:24 ceridwen joined #marpa
13:48 rns koo7: you can use markers -- https://metacpan.org/pod/distribut​ion/Marpa-R2/pod/Event.pod#Markers e.g. https://gist.github.com/rns/962fdb4f30d0681cc07d
13:49 rns e.g. if you define then ::= 't' t_marker 'h' th_marker 'e' the_marker 'n' then_marker
13:50 rns then you can handle nulled events on *_marker symbols and see that the parse is in then clause.
13:51 rns in the above gist I used markers to catch minus signs which are not wrapped into whitespaces (awkward minuses)
13:57 rns the request was 'to test for mandatory whitespace' in a comment to an SO question -- http://stackoverflow.com/questions/27109840/m​arpa-can-i-explicitly-disallow-keywords-as-id​entifiers/27120210#comment48471794_27120210
16:13 teatime joined #marpa
16:47 koo7 joined #marpa
17:22 idiosyncrat_ joined #marpa
17:25 idiosyncrat_ rns: re http://irclog.perlgeek.de/m​arpa/2016-04-05#i_12287271 -- thanks, missed that one.
18:24 idiosyncrat_ Next steps with Marpa::R3 --
18:25 idiosyncrat_ The next goal is to get ready to refactor, eliminating the NAIF.
18:25 idiosyncrat_ But before we can eliminate the NAIF, we have to eliminate all its test from the test suite.
18:26 idiosyncrat_ Most can simply be deleted, but some don't have duplicates using the SLIF and these SLIF duplicates should be created.
18:26 idiosyncrat_ So I'm looking for volunteers.
18:26 idiosyncrat_ Two I'
18:27 idiosyncrat_ Two I've identified as needing converstion from NAIF to SLIF
18:27 idiosyncrat_ cpan/t/jirotka.t
18:27 idiosyncrat_ cpan/t/catalan.t
18:28 idiosyncrat_ I'm currently censusing the tests, identifying which ones can be deleted, which do not use the NAIF, and which need to be converted.
18:37 idiosyncrat_ As a note, these tests contain tests of the NAIF diagnostics -- where there is an equivalent documented for the SLIF, one should be substituted for the other ...
18:38 idiosyncrat_ Where there is no SLIF equivalent, the NAIF diagnostic can be simply deleted.
18:49 rns idiosyncrat: jirotka.t has tests for show_ahms() grammar method and show_earley_sets(), show_and_nodes(), show_or_nodes() recognizer methods which AFAIK have no equivalents in SLIF.
18:51 idiosyncrat_ Actually, rethinking, I think we want to keep the tests even in there is an *undocumented* SLIF equivalent.
18:51 idiosyncrat_ Let me look at the code re the ones you mention
18:51 rns ok
18:53 idiosyncrat_ I think I'll "wrapper" those tests so that they are undocumented SLIF tests, and we can keep them.
18:53 idiosyncrat_ Unless you want to try to take a crack at doing that.
19:00 idiosyncrat_ jdurand: I *do* remember your request to have rule alternatives, when they are of equal precedence in a precedenced rules, be automatically ranked in lexical order, with the first having the highest rank.
19:01 rns idiosyncrat: Well, I think I can do sl_jirotka.t and sl_catalan.t which will test only documented SLIF equivalents.
19:01 idiosyncrat_ I have in mind how to do that, but I want to defer that until *after* we convert these tests, because I may well want to get some refactoring in while adding that feature.
19:01 idiosyncrat_ rns: Why not hold off until I wrapper all the diagnostic methods?
19:02 idiosyncrat_ I should be able to do that in the next 24 hours.
19:02 idiosyncrat_ If you could list the ones I'll need to do, that would be immensely helpful.
19:04 rns Ok, here goes -- jirotka.t: show_ahms() NAIF grammar method and show_earley_sets(), show_and_nodes(), show_or_nodes() NAIF recognizer methods
19:05 rns catalan.t seems to use no diagnostic methods.
19:06 rns Holding off looks good -- I won't be able to take jirotka.t and catalan.t until this Friday and weekend.
19:08 rns Once you have them wrappered, I'll move all documented SLIF equivalents to sl_jirotka.t and sl_catalan.t leaving the above listed in jirotka.t and removing catalan.t, if I am understanding the task right?
19:23 idiosyncrat_ rns: Thanks
19:23 idiosyncrat_ Leave *both* the original jirotka.t and catalan.t in place
19:24 idiosyncrat_ But, yes, name the new versions sl_jirotka.t and sl_catalan.t
19:25 rns ok, will take it and let you know.
20:13 Pursuit joined #marpa
22:26 koo7 joined #marpa
23:56 idiosyncrat_ joined #marpa

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