Camelia, the Perl 6 bug

IRC log for #ironperl, 2009-02-23

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

All times shown according to UTC.

Time Nick Message
17:38 cj lalala
17:41 diakopter hi
17:43 diakopter so, I need to start solidifying the plan.  I've done a ton of discovery, and it's paid off (it seems).  There exist a ton of open-source compiler/interpreters targeting .Net from which to draw ideas, namely Microsoft's CSharp and FSharp compilers.  Microsoft's CSharp one will obviously be the easiest for me to understand.
17:43 diakopter I'll refer to it as CCI from now on
17:43 cj okay
17:44 cj did you get word whether it was mistakenly published?
17:44 cj (Sara is at Mardi Gras)
17:44 diakopter no reply from the other two I CCed, no
17:50 diakopter so, CCI's *API* will be the initial target of the bootstrap.  No need to create another runtime layer or type system abstraction like I was envisioning.
17:50 diakopter just... disable type-checking and create types as needed on the fly.
17:52 diakopter omg.
17:52 diakopter lazy lists in C#, auto-parallelization in C#,
17:55 cj var foo = new Some.Complicated.Blahblah.Type();
17:55 cj that's my favorite new piece
17:55 diakopter yeah; var
17:55 diakopter here's a paper on implementating continuations on MSIL/CLR
17:55 diakopter that should be "easy" to shoehorn into CCI ;)
17:56 cj you know the linden labs folks have a (crappy) continuation implementation, right?
17:56 diakopter no... /me goes searching
17:56 cj it's internal, but you might be able to pry it out
17:56 cj #opensl on uhrm... efnet?
17:56 * cj checks
17:57 diakopter oh, here's your blog entry onit
17:57 cj I'd say #lsl is a good place to talk
17:58 cj sorry for the terribly slow web server thing
17:59 diakopter so can you attend lang.net09?
18:01 cj yeah, I've gotten the go-ahead from my manager
18:01 cj now all I need to do is get an okay from the wife :)
18:02 diakopter don't you live in the area?
18:03 cj yeah.  I even have a ride to campus :)
18:03 cj but I won't be paid to attend.
18:04 diakopter ah, ok. I'll be getting a plane ticket this week.  I guess I don't need a rental car with the nice bus system?
18:04 cj where are you staying?
18:07 diakopter in a rental car if I get one? at a national park?  I'm cheap.
18:11 cj heh, we have a couch.
18:15 cj (which you're welcome to crash on)
18:15 cj in which case you'd have a room and a ride
18:15 cj though we leave between 1630 and 1730 usually, which would be before the nightly festivities.
18:16 cj there are busses, but it's less than fun to get to Everett from Redmond on the bus.
18:16 cj er, Redmond to Everett
18:16 diakopter I don't do festivities, generally.
18:17 diakopter thanks for the couch offer; sounds great.
18:17 diakopter oh; you're in Everett?
18:18 cj a'yup.  I commute to Redmond daily.
18:27 cj Shri responded
18:35 diakopter ooooooo\
18:36 diakopter here, I'll draft my reply here
18:37 cj okay
18:37 cj I'll nod like I understand
18:38 diakopter well, continuations are a perfect example
18:38 diakopter some language features of Perl 6 would be far easier to implement at the C# compiler level
18:39 diakopter (generating certain non-C# MSIL while compiling the Perl6Compiler
18:39 diakopter )
18:40 cj why not use the mono c# compiler instead of the CCI?
18:41 diakopter I haven't explored the source of mcs; I suppose it's possible that it's very similar, if not derivative.
18:41 cj probably not derivative, no
18:41 cj marek is your guy for mcs, btw
18:41 cj he's very responsive, too
18:42 cj he's fixed at least 3 bugs I've brought up while fiddling with ironruby
18:43 diakopter I guess my first inclination is to use it because I have a stronger natural trust for it, since it's been around a lot longer and is used by msft.  I mean, I realize there's a regression test suite for mcs, but (I'm guessing entirely here) it's probably tiny compared to Microsoft's test suites for its C# compilers
18:45 diakopter I mean, I don't intend to diss the mono project at all or anything related to it.  it's just a lot more resources are/were behind Cci... or so I'm assuming.
18:47 cj yeah, I haven't looked at the regression suite on the msft side.  the mono side is pretty impressive from what I've seen, though :)
18:50 diakopter oh
18:51 diakopter well, perhaps I'm mistaken.
18:53 cj cjac@dev0:/usr/src/svn/mono/mcs$ find tests  -name '*.cs' | wc -l
18:53 cj 1620
18:56 cj all bugs filed and work done gets a new test
18:56 cj that 'test driven development' stuff
19:05 diakopter yeah

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