Perl 6 - the future is here, just unevenly distributed

IRC log for #perl11, 2015-04-08

| Channels | #perl11 index | Today | | Search | Google Search | Plain-Text | summary

All times shown according to UTC.

Time Nick Message
00:39 willthechill joined #perl11
01:47 ilbot2 joined #perl11
01:47 Topic for #perl11 is now Perl5 + Perl6 == Perl11; # http://irclog.perlgeek.de/perl11/
07:40 rurban_ joined #perl11
07:57 travis-ci RPerl build passed. Will Braswell says 'Grammar, AST to Class Mapping, Part 1'
07:57 travis-ci http://travis-ci.org/wbraswell/rperl/builds/57605032 https://github.com/wbraswell/rperl/compare/458994ffde76...9ccb441cdca8
08:13 basiliscos joined #perl11
18:00 rurban_ joined #perl11
19:02 willthechill cool
19:03 willthechill rurban_: RPerl v1.0 grammar has been frozen, apps may now be written with confidence, starting on final piece which is AST to C++ code generator
19:42 rurban_ great
19:42 rurban_ I still have to check if I like your names though
19:42 rurban_ Because I'm preferring to go for perl6 established names, esp. types
19:43 willthechill haha well if you have a problem with RPerl types then it is TRIVIAL to add type aliases
19:44 willthechill dozens of array subtypes here to show you how easy type aliasing is:  https://github.com/wbraswell/rperl/blob/master/lib/RPerl/DataStructure/Array/SubTypes.pm
19:44 willthechill my types follow C++ and Perl 5 where applicable
19:45 rurban_ yeah, I'm still not sure. will see when my new type inferencer will start working.
19:46 willthechill dynamic types are BAD, mmm-kay?  :-)
19:46 rurban_ I'll go for shorter names: int, num, str, bool
19:46 willthechill short name aliases can be added with no effort
19:47 rurban_ for the type inferencer aliases are probably a bigger deal.
19:48 rurban_ But I probably provide read-only core versions (int, num, str) and user-overridable versions as in perl6: Int, Str, Num
19:49 rurban_ I also have the undef variants with added ? suffix
19:49 rurban_ int? = int|undef
19:50 willthechill well I'm glad you are the one working on dynamic types and Perl 6 types because I sure as heck am not interested in it!
19:50 rurban_ no name for the magic variants, they are probably just scalar/array/hash
19:50 rurban_ yes, it's a nightmare
19:50 willthechill we have "scalartype" in RPerl because "scalar" is already protected Perl 5 word
19:50 rurban_ but typed arrays and typed ints make it really fast
19:51 rurban_ just inference cannot really do much without typed signatures in most modules. and this is boycotted still upstream
19:52 willthechill dynamic types are the devil
19:52 willthechill static types are from heaven
19:52 willthechill THOU SHALT NOT USE DYNAMIC TYPES!
19:52 rurban_ And I used attributes for type variant syntax which is its own nightmare
19:52 willthechill hahaha
19:53 rurban_ but it's getting more and more static. some of the old dynamic modules are already static.
19:53 willthechill well that is good news!
19:53 rurban_ dynaloader, xsloader, strict, attributes, carp, ...
19:53 rurban_ warnings, config, ...
19:54 rurban_ unicode next
19:54 willthechill BTW I can be available for Germany Perl workshop but I would need airfare scholarship...
19:54 rurban_ it's not so important and we have no budget
19:55 rurban_ big announcement will be this fall I hope
19:55 willthechill big announcement of Perl 6?
19:56 rurban_ no
19:56 rurban_ biggest problems is getting names (GVs) static
19:56 willthechill so big announcement of static types in Perl 5?
19:57 rurban_ this and a bit more
19:57 rurban_ I hope to get a mop also in. class, method, role
19:58 willthechill oh no why MOP in Perl 5?  it just makes it MORE SLOW!!!
19:58 rurban_ roles are staticallly composable classes.. .faster, not slower
19:59 willthechill oh well faster is better
19:59 rurban_ And we need to optimize method calls, much faster
19:59 willthechill but still for me, any interpretation or dynamic behavior is too slow
19:59 rurban_ when we can close classes it's a new world
20:00 rurban_ like in your world
20:00 willthechill but you are correct, I'm glad we will have faster Perl 5 interpreter
20:01 rurban_ and the compiled versions will be undestroyable by mockers, devel::cover, test mocking
20:02 rurban_ most of cPanel binaries currently fail when being devel::cover'ed
20:02 willthechill well that sounds good
20:03 rurban_ this affects interaction with compiled shared libs. B::C, B::CC or rperl
20:03 rurban_ currently it's still too dynamic. I have to lock it down more, more to rperl or B::CC
20:04 rurban_ mockers should compile their own versions. unit testing or coverage testing
20:07 rurban_ the new type checker and inferencer is complety compile-time btw. Written in C.
20:08 rurban_ in perl it needs too much memory and it was too slow

| Channels | #perl11 index | Today | | Search | Google Search | Plain-Text | summary