Perl 6 - the future is here, just unevenly distributed

IRC log for #pdl, 2014-05-23

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

All times shown according to UTC.

Time Nick Message
00:00 mohawk run4flat, round tuits for what?
00:00 * mohawk evinces caution
00:03 mohawk C::Block - for when XS is just too fucking simple and you need MORE COMPLEXITY in your life!
00:03 * mohawk starts cutting himself
12:50 chm joined #pdl
12:51 chm run4flat: Extremely cool stuff.  I'm looking forward to trying this out.  PP+JIT would be great for PDL3
14:12 run4flat joined #pdl
15:03 chm joined #pdl
15:04 chm run4flat: This tcc work is just the piece to tie together the PDL3 work.
15:04 chm It is interesting that the Prima object system and the ongoing perl5-mop work are very similar
15:04 run4flat chm, I hope so
15:04 run4flat :-)
15:04 chm I would like to see the PDL3 compute kernel take advantage of both to streamline things
15:05 run4flat as for the Prima and p5-mop stuff, I hadn't looked closely
15:05 run4flat chm, that would be awesome. :-)
15:05 chm Steven Little had a blog post on some of the details.
15:05 run4flat I read what he wrote about a week ago
15:05 chm I realized that my previous thoughts about PDL3 implemented as a C-accessible layer
15:06 chm that ties in with perl seamlessly was pretty much what Prima had started and
15:06 chm then I saw the posts from Steven L and realized he was heading in a similar direction.
15:06 chm Your C::Block stuff is right down that road---and shows promise to clean up the implementation greatly!
15:07 run4flat I've thought a lot about a C-based object system that supports roles
15:07 run4flat it's a bit tricky
15:07 run4flat Prima on its own does not support roles at the C level
15:07 run4flat it might be able to use it to implement roles at the Perl level, though
15:07 chm I think it should be doable---in fact, it is just a dynamic version of what Prima already does, isn't it?
15:07 run4flat but yes, I hope that C::Blocks can bring a lot fo things together
15:07 run4flat not quite
15:08 run4flat you need to be able to quickly look up role functions by struct member name
15:08 run4flat and you can't be guaranteed a struct layout with roles
15:08 run4flat so I think it would require an additional role-based vtable
15:08 chm I guess I'm confused.  The whole point of a role is that it is folded *into* the composed class.
15:09 run4flat For exampe, imagine that we want to have an "ArrayLike" role
15:09 chm You only have to do that once and then it is the same as before...
15:09 run4flat you want to use a PDL-like back-end for one implementation, and Perl-array-like backend for another
15:10 chm I'm not sure I follow?
15:10 * run4flat thinks he needs to rethink his thinking
15:10 run4flat :-)
15:10 run4flat lemme think more about it
15:10 chm OK,
15:11 chm I think that the thing that may be confusing is the ability to apply a role to an object
15:11 run4flat no...
15:11 run4flat OK, I got it
15:11 chm The key there is that a new class is instantiated with the role composed into that and then an object of *that* class returns.
15:12 run4flat So, from the C level, you'd be constantly wrapping and re-wrapping your objects in different role wrappers?
15:12 run4flat I'm just thinking about struct layouts
15:12 chm Only if you are applying new roles---right?
15:12 run4flat hmm
15:12 chm Do use cases exist for removing roles?  I don't know of any.
15:13 run4flat I think we'd have to sit down at a table with a big sheet of paper to understand each other here. :-)
15:13 run4flat no, not removing them
15:13 chm I would love to.  When are you visiting your family here?
15:13 run4flat but simply writing code that knows, "Hey, this thing that you will send me (later), it has this role"
15:13 run4flat chm, I don't know when I'll be in DC next
15:14 run4flat speaking of which---I'm sorry I missed the hackathon at DCBPW this year
15:14 run4flat :-(
15:14 run4flat but I'm sure I'll come down this summer
15:14 chm OK, I was hoping to push things so you would have some PDL2 + 64bit support for you Orlando excursion
15:14 run4flat alas, I'm not going to that either. :-(
15:15 chm Unfortunately, my windows / cygwin development environments are enough clunky that I've been stuck.
15:15 run4flat I'll be in College Park for a New PHysics Teacher thing
15:15 chm Oh?  Thought you were.
15:15 run4flat I had planned on it
15:15 chm Ok.
15:15 run4flat I had promised to be on a panel.
15:15 chm Well, I'm sorry to run but lunch calls.  Great to see you back again.  Will try to chat more soon.  o>
15:16 chm o/ (I mean)
15:16 run4flat o/
15:43 run4flat joined #pdl
16:27 mohawk joined #pdl
18:06 drrho joined #pdl
18:10 run4flat joined #pdl
21:35 run4flat joined #pdl
22:48 run4flat joined #pdl

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