Perl 6 - the future is here, just unevenly distributed

IRC log for #marpa, 2016-06-27

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

All times shown according to UTC.

Time Nick Message
00:07 idiosyncrat_ joined #marpa
02:34 ronsavage http://phys.org/news/2016-06-b​irds-abstract-grammatical.html
05:07 idiosyncrat_ Good night!
06:50 ronsavage joined #marpa
08:01 black_ant joined #marpa
10:07 rns joined #marpa
10:11 shadowpaste "rns" at 217.168.150.38 pasted "test failures at commit 269fe94 under cygwin (gcc 5.3.0, perl 5.22.2), last t+ commit: 9c46219" (1991 lines) at http://fpaste.scsys.co.uk/524735
10:11 rns idiosyncrat: the above boils down to
10:11 rns Bad op code (6, 'unknown') in v->stack_step, step_type 'MARPA_STEP_TOKEN' at /home/Ruslan/Marpa--R3/cpan​/blib/lib/Marpa/R3/Value.pm line 1426.
10:13 rns AFK, will backlog.
10:13 rns left #marpa
10:45 kaare_ joined #marpa
12:41 rgrinberg joined #marpa
13:35 black_ant joined #marpa
14:37 black_ant joined #marpa
14:56 idiosyncrat_ joined #marpa
14:57 idiosyncrat_ Uploaded Marpa-R3-4.001_014 last night.
14:58 idiosyncrat_ I is one of those very annoying releases.  It succeeds 100% on my test boxes, but fails literally everywhere else.
15:06 idiosyncrat_ rns: could you bisect the problem?
15:07 idiosyncrat_ I've made a point of using smaller commits since the last issue.
15:13 rns joined #marpa
15:14 rns idiosyncrat: well, I did, sort of -- it starts failing at commit 269fe94, last good commit being 9c46219
15:17 idiosyncrat_ rns: Thanks -- I'm about to leave for an appointment, so won't be able to look at it until this afternoon.
15:21 rns and, with directory separator changed, it pass all tests under windows, so it can be smth. similar to Marpa-R3_04.001_011
15:24 idiosyncrat_ "similar to Marpa-R3_04.001_011" ?
15:32 rns yes, that it also failed everywhere except windows and your dev boxes and was fixed by this PR -- https://github.com/jeffreykegler/Marpa--R3/pull/29
15:33 rns not sure how it's related to changes introduced in commit 269fe94 though
15:36 rns Marpa-R3_04.001_011 -- http://matrix.cpantesters.or​g/?dist=Marpa-R3%204.001_011
15:36 rns http://irclog.perlgeek.de/m​arpa/2016-06-11#i_12649033
15:50 idiosyncrat_ rns: if you can find anything more about the current problem(s), it'd be greatly appreciated.  Thanks!
15:50 idiosyncrat_ I have to leave for an appt now.
15:50 idiosyncrat_ AFK
17:10 rgrinberg joined #marpa
17:13 rns idiosyncrat: looks like p_ops_sv doesn't get proper value after https://github.com/jeffreykegler/Marp​a--R3/blob/master/cpan/xs/R3.xs#L1792 before https://github.com/jeffreykegler/Marp​a--R3/blob/master/cpan/xs/R3.xs#L1796
17:25 rns marpa_v_token (v) returns 28  and  SV **p_ops_sv = av_fetch (v_wrapper->token_semantics, marpa_v_token (v), 0); returns 0
17:30 rns 28 in t/01-lua.t that is
17:34 black_ant joined #marpa
17:51 shadowpaste "rns" at 217.168.150.38 pasted "diagnostics of Marpa-R3_04.001_014 test failures per the above" (2396 lines) at http://fpaste.scsys.co.uk/524762
18:02 shadowpaste "rns" at 217.168.150.38 pasted "01-lua.t under windows (success)" (57 lines) at http://fpaste.scsys.co.uk/524763
18:03 shadowpaste "rns" at 217.168.150.38 pasted "01-lua.t under cygwin (failure)" (12 lines) at http://fpaste.scsys.co.uk/524764
18:06 rns failure under cygwin in 01-lua.t is at do_recce_test($salve, [], ['salve, munde!'], 'Salve, 0 args');
18:20 rgrinberg joined #marpa
18:23 rns based on 'Looks like your test exited with 255 just after 17.' in the cygwin log above
18:41 rns which frankly doesn't make sense because do_recce_test() doesn't go get in v_do_stack_ops() where the error occurs.
18:41 rns that's about it so far -- late here, AFK, will backlog.
18:41 rns left #marpa
19:59 choroba joined #marpa
20:02 choroba Hi, I'm trying to implement a new feature to my "compiler", but with no success so far. Is it possible to add positions in the input string to the structure produced by the parser?
20:02 choroba I'd like to "debug" the parsed structure, i.e. show where different parts came from
20:04 choroba I thought events might help me, but I'm not able to interlink events with actions
20:05 rns joined #marpa
20:08 rns choroba: there are start and length array descriptors, which you can add to value array -- https://metacpan.org/pod/distribution/Marpa-R​2/pod/Semantics.pod#Array-descriptor-actions
20:08 rns e.g.
20:08 rns :default ::= action => [ name, start, length, values ]
20:08 rns lexeme default = action => [ name, start, length, value ] latm => 1
20:09 rns that way, each AST node will have the input span
20:09 rns corresponding to the parsed symbol
20:10 choroba oh, I've read the document, but somehow skipped this section
20:10 choroba *blush*
20:11 choroba Thank you, that's exactly what I need
20:12 rns great!
20:12 rns .part
20:12 rns left #marpa
22:15 rgrinberg joined #marpa
22:20 rgrinberg joined #marpa
22:21 idiosyncrat_ joined #marpa
23:17 ronsavage joined #marpa
23:41 idiosyncrat_ rns: could you try testing with commit 7a5ae5df7bc07054cd39b49b669b3542d85d4c4f on the master branch.
23:42 idiosyncrat_ It fixes a dangerous type mismatch, which could easily be disastrous on architectures where the sizes of an XS UV and a C int differed.
23:44 idiosyncrat_ And the mismatch was introduced in commit 269fe94, the one pointed to by your bisection.

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