Perl 6 - the future is here, just unevenly distributed

IRC log for #marpa, 2016-04-27

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

All times shown according to UTC.

Time Nick Message
01:37 ronsavage Re Marpa-R3-4.001_002. All tests passed, and it installed, on my Deb box.
02:54 rns joined #marpa
03:20 idiosyncrat_ ronsavage: Thanks!
04:01 rns idiosyncrat: re http://irclog.perlgeek.de/marpa/2016-04-26#i_12397607 -- yes, no regressions  and even one netbsd pass
04:01 rns [2016-04-27T03:48:12Z] [Chris Williams (BINGOS)] [pass] [JKEGL/Marpa-R3-4.001_002.tar.gz] [amd64-netbsd-thread-multi] [perl-v5.12.2] [ddb75b00-0c2a-11e6-bd2f-67968fb2e322] [2016-04-27T03:48:12Z]
04:02 idiosyncrat_ rns: what data are you looking at?
04:03 rns http://metabase.cpantesters.org/tail/log.txt
04:05 idiosyncrat_ Thanks, I'd be trying to remember where that page is.
04:06 idiosyncrat_ I'd been using http://matrix.cpantesters.org/?dist=Marpa-R3+4.001_002 which is maybe 24 hours delayed
04:07 idiosyncrat_ One netbsd PASS is a very good sign, because previously we'd been 10% FAIL on netbsd, so it's probably a sign we've fixed something if not everything.
04:11 rns Yep. Eliminating stamp-h1 looked like fixing it and this netbsd pass gives stronger hopes.
04:12 rns BTW log.txt only has IIRC 1000 or so last reports so can/needs to be checked relatively frequently.
04:29 rns idiosyncrat, ronsavage: re sub action () { my ($ppv, $array_ref) = @_; -- looks good: eliminates a special case and allows accessing child0 as $array_ref->[0] or $_[1]->[0]
04:29 rns BTW, now that [ name, values ] is default semantics (and given that name adverb allows setting fancy names to rule alternatives, the default action sub can be written as
04:29 rns sub action { my ($ppv, $name, @values) = @_; # ... dispatching code based on $name ... }
04:29 rns idiosyncrat: re blessing -- well, FWIW, it allows implementing the Interpreter pattern -- one doit() method in many packages -- pity bless => ::name is not allowed for RHS alternatives -- I can see why it is not, unlike ::lhs, but still.
04:31 idiosyncrat_ rns: Glad to hear it, because it makes what I want to do with the evaluation internals easier if the result of an evaluation is *always* an array.
04:32 idiosyncrat_ I'm in no hurry to eliminate blessing, and there's no need to if there is some use case for it.
04:32 idiosyncrat_ And we can even look at extending it.
04:33 idiosyncrat_ extending it to bless => ::name, that is.
04:34 idiosyncrat_ It's been a while so I'm not 100% sure why that's not allowed -- it may be because development order was
04:34 idiosyncrat_ 1.) bless
04:34 idiosyncrat_ 2.) rule names
04:35 idiosyncrat_ and I never bothered going back to bless and extending it.
04:37 rns Well, I thought that's because ::lhs (which comes from the grammar source), can, unlike ::name (which is user-specified), be guaranteed to have no chars not alllowed in package names.
04:38 idiosyncrat_ But I could work around that very simply by checking blessings, and if I don't like any name characters, throwing an exception.
04:38 idiosyncrat_ That leaves it up to the user to pick less ornate names, or give up on using them as blessings, his choice.
04:40 rns Just looked at http://jeffreykegler.github.io/Ocean-of-Awareness-blog/individual/2013/03/bnf_to_ast.html
04:41 idiosyncrat_ Yes, that was something folks dreamed about back in the 70's.
04:42 idiosyncrat_ It was interesting to announce it in the 2010's and have it received with all the enthusiasm of a sale on lava lamps and bell-bottom jeans.
04:42 rns :)
04:43 rns ... there is bless => and -- blessing to a Perl word (ad documented) -- that sort of alleviates the needed to have bless => ::name
04:43 rns s/ad documented/as documented/
04:45 idiosyncrat_ Once the refactoring is done, we can at least glance at a lot of possible changes.
04:45 idiosyncrat_ At this point I don't mind breaking compatibility, which makes this the time to do it.
04:46 idiosyncrat_ It's late CA time -- good night, will backlog as usual.
04:47 rns Yep there is always Marpa::R2 for compatibility -- good night.
05:01 rns idiosyncrat: re cpantesters on freebsd
05:01 rns cc  -shared  -L/usr/local/lib -fstack-protector   -o blib/arch/auto/Marpa/R3/R3.so xs/R3.o xs/libmarpa.a
05:02 rns cc: xs/R3.o: No such file or directory
05:02 rns in http://www.cpantesters.org/cpan/report/3be852a0-0be1-11e6-b39d-b795fcd2507e among others
05:02 rns looks related to MYEXTLIB => 'xs/R3$(OBJ_EXT) xs/libmarpa$(LIB_EXT)',
05:02 rns in Makefile.PL
05:06 rns re make[1]: don't know how to make test_dynamic. Stop in http://www.cpantesters.org/cpan/report/1d4134a8-0be0-11e6-a0cf-316efcd2507e
05:07 rns perhaps explicitly SKIP      => [qw(test dynamic dynamic_lib dlsyms)] as in xs/Makefile.PL
05:08 rns AFK, will backlog.
05:08 rns left #marpa
05:58 ronsavage joined #marpa
07:01 beaugunderson joined #marpa
07:08 iarna joined #marpa
08:21 maybekoo2 joined #marpa
10:12 kaare_ joined #marpa
14:20 idiosyncrat_ joined #marpa
14:26 shadowpaste joined #marpa
15:51 maybekoo2 joined #marpa
16:17 JPGainsborough joined #marpa
17:26 choroba joined #marpa
19:19 idiosyncrat_ Errands to run -- AFK
22:22 ronsavage rns: Re http://irclog.perlgeek.de/marpa/2016-04-27#i_12398292. I always use $$array_ref[0], since I reserve -> for method calls. The fact that it's 1 char shorter does not matter.
22:49 idiosyncrat_ joined #marpa
22:53 idiosyncrat_ ronsavage: I also use that form but, following a "best practice" from Damien's book, I always write it as
22:53 idiosyncrat_ ${$array_ref[0]}
22:53 idiosyncrat_ for clarity
22:56 ronsavage I think the suggestion from Damien on that point just adds {} without adding value or clarity.
23:09 choroba ${$array_ref[0]} isn't the same as $$array_ref[0]
23:11 choroba ${$array_ref}[0] is possible, though
23:26 idiosyncrat_ choroba: Oops.  Shows why I always add the braces.
23:26 idiosyncrat_ :-)

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