Perl 6 - the future is here, just unevenly distributed

IRC log for #marpa, 2016-04-26

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

All times shown according to UTC.

Time Nick Message
02:07 idiosyncrat_ I've just uploaded a new trial version of R3 to CPAN: Marpa-R3-4.001_002.tar.gz
02:07 idiosyncrat_ It includes all my recent refactorings, but also some changes to the build logic
02:08 idiosyncrat_ I've stopped using stamp-h1 in the build directories -- the supposed portability advantage of using a timestamp file never emerged.
02:11 idiosyncrat_ And I've rearranged the dependencies to be more direct and straightforward.  I can't think of a reason the old ones would fail on BSD for non-parallel makes, but in any case the current ones might even succeed on parallel makes -- though I have not tested that.
02:12 idiosyncrat_ But the new way of arranging the dependencies is more straighforward, and if we have to get into a lot of build/Makefile debugging, they're a better place to start.
02:12 idiosyncrat_ Hopefully there are no regressions.
02:13 idiosyncrat_ Btw, the Raspberry Pi 3 is faster, but there is the standard problem with multicore machines -- of its 4 cores, 3 are almost never used.
02:58 ronsavage joined #marpa
04:21 idiosyncrat_ Good night!
07:03 ronsavage joined #marpa
07:46 maybekoo2 joined #marpa
12:15 kaare_ joined #marpa
19:22 choroba joined #marpa
19:57 sivoais joined #marpa
21:30 maybekoo2 joined #marpa
22:23 idiosyncrat_ joined #marpa
22:59 idiosyncrat_ Wrt Marpa::R3, I'm looking at the semantics, and right now a rule ...
23:00 idiosyncrat_ usually sees its child value in its arguments as a list ...
23:01 idiosyncrat_ sub action () { my ($ppv, $child0, $child2, @other_children) = @_; ...
23:02 idiosyncrat_ were $ppv is the per-parse-value, $child0 is the value of the first child etc.
23:02 idiosyncrat_ But you can ask for the child values to be put into an array with the ::array reserved action name *AND* ...
23:03 idiosyncrat_ if you asked for the rule alternative to be "blessed", the child values go into an array, whether you ask for it explicitly or not.
23:05 idiosyncrat_ This last is because to bless the arguments to a rule, there has to be something "bless-able" -- Perl arrays are single objects and can be blessed, but Perl lists are not single objects in this sense, and cannot be blessed.
23:05 idiosyncrat_ All of which is to say the behavior is not terribly orthogonal.
23:06 idiosyncrat_ It's the way it is for historical reasons .. original putting the child values into the arguments as a list seemed the natural thing to do, but then I added the feature of "blessing" rule alternative and had to create the special case.
23:07 * idiosyncrat_ prepares to ask a question of those still following him.
23:07 ronsavage So are you going to make this the behaviour: sub action () { my ($ppv, $array_ref) = @_;
23:08 idiosyncrat_ That was going to be my question, yes.
23:08 idiosyncrat_ Right now, it is that way sometimes and sometimes not.
23:08 idiosyncrat_ My question is should it make it so that it is always the case that
23:09 idiosyncrat_ sub action () { my ($ppv, $array_ref) = @_;
23:09 idiosyncrat_ or
23:09 ronsavage I for one would prefer the $array_ref approach, so there would always only be 2 params. In practice I don't like the current way - it's confusing.
23:10 idiosyncrat_ OK, that's important feedback.
23:10 idiosyncrat_ While we're on the topic, I'd also like to know if the "blessing" feature is important to anyone.
23:10 idiosyncrat_ In my own experiments I found it not as helpful as I'd hoped.
23:19 ronsavage I don't think I've ever used it, but I can imagine it could be of interest to some people.
23:20 ronsavage AFK
23:34 idiosyncrat_ rns: thanks for the tests of Marpa-R3-4.001_002 -- at least there was no regression :-)
23:47 ronsavage joined #marpa

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