Perl 6 - the future is here, just unevenly distributed

IRC log for #pdl, 2014-06-20

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

All times shown according to UTC.

Time Nick Message
10:39 chm joined #pdl
10:43 chm mohawk: An image piddle with shape [4 200 100] has data organized by pixels (rgba)  in order along the rows (x) and rows repeat until image is complete (y)
10:44 chm mohawk: A [200 100 4] piddle corresponds to a sequence of per color images (think byte-planes)
10:55 chm sivoais, perigrin, mohawk: I would think AnyEvent would be the way to go: extremely stable, flexible, has perl-only backend, works with other/multiple event loops...
10:57 chm sivoais, perigrin, mohawk: The weaknesses for windows platforms are well characterized and documented.  Have you used AnyEvent and had a problem?
13:01 vicash sivoais: since AnyEvent supports multiple backends, you can ask ANyEvent to use POE on Windows and use EV on Linux/BSD to allow for safe runs. Although I do think EV which is also written by Marc works fine on Windows except for the fork() events
13:02 vicash Prima on Windows also uses POE internally and I have used Prima with both POE and AnyEvent and support for POE on Windows is decent enough
13:49 drrho joined #pdl
14:01 mohawk joined #pdl
16:13 sivoais IO::Async supports all those backends too... POE, EV, Mojo, even select(2)
18:27 chm joined #pdl
18:33 chm vicash: I don't see any use of POE by Prima in the source.
18:35 chm sivoais: Whichever event framework is selected, the key is to ensure that features used work on windows.
18:39 sivoais yes, definitely
18:40 sivoais chm: just submitted this report <https://sourceforge.net/p/pdl/feature-requests/82/>
18:40 sivoais I couldn't find discussion on this behaviour on the mailing list archives
18:41 sivoais I'm going to step out for a bit, but I'll be back in an hour or so
18:43 sivoais oh, and I'm almost done with the alpha version of this IPython kernel. I need to refactor it and add tests!
18:47 chm sivoais: FWIW, AnyEvent, EV, and IO::Async all installed out of the box on cygwin with cpan.
18:48 chm sivoais: AnyEvent can use IO::Async as a backend, and IO::Async supports using AnyEvent so they are matched there.
18:49 chm sivoais: However, a selling point for AnyEvent is that it doesn't lock a user of a module into a specific event loop.
18:50 chm sivoais: The README for IO::Async says it is still in development---don't know if that means the API is still changing.
18:51 chm sivoais: AnyEvent seems to have a lot of edge case fixes for various platforms and backends.
18:52 chm sivoais: If iPerl is not planned to have a module that could be used, it probably doesn't make a difference which is used.
18:52 chm sivoais: My preference would be AnyEvent for the flexibility, maturity, and the fact that I'm more familiar with it :-)
19:04 chm vicash: POE failed to pass tests on a cpan install.  Don't know the origin of the problem---tweaking may get things to work.
19:21 vicash chm: you have to install POE::Loop::Prima for Prima to use POE. and then include "use POE 'Loop::Prima';' to get Prima to use it
19:22 vicash chm: on strawberry perl i had no problem with POE but i dont know about cygwin
19:40 sivoais I'll take a look at AnyEvent in a bit. I've used all three (POE, AnyEvent, IO::Async) and they are just different APIs to the same thing
19:40 sivoais really, all I need for the loop is a sub run {} and let it just route messages around :-P
20:01 vicash i would keep it as simple as possible unless necessary.
23:01 perigrin IO::Async has to be used with AnyEvent in the approved way
23:02 perigrin https://metacpan.org/source/MLEHMANN​/AnyEvent-7.07/lib/AnyEvent.pm#L1412
23:02 perigrin otherwise AnyEvent will have an issue.

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