Camelia, the Perl 6 bug

IRC log for #ironperl, 2009-03-03

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

All times shown according to UTC.

Time Nick Message
16:00 diakopter cj: ping
16:09 cj huh?  wha?
16:10 diakopter hi
16:10 diakopter what do you know about VSTA
16:10 diakopter visual studio tools for applications
16:10 diakopter ... used in infopath, sql server
16:10 cj nothing.  but I know people who do...
16:10 diakopter ok
16:11 cj got any specific questions?  I can send an intro email after finding a good contact.
16:11 diakopter on its pricing/licensing page (its sales/support are managed by SummitSoft) it implies that you can redistribute it for free
16:11 diakopter its tools would be AWESOME for me...
16:11 cj you still planning to present at Lang.NET?
16:12 diakopter it seems to be the evolution of what used to be VSA
16:12 diakopter yeah, I hope so
16:12 diakopter (if they let me)
16:12 diakopter but I'm attending regardless.... :P
16:12 diakopter I don't need a 15-min speaking slot to disseminate news about what I'm working on :P
16:13 diakopter (necessarily)
16:13 diakopter though a whizbang presentation wouldn't hurt
16:13 diakopter why do you ask?
16:15 cj I figure if I tell my friend that you're presenting at a MS get-together, he'll be more likely to help you out :)
16:17 diakopter ah :)  ftp://ftp.summsoft.com/vsta/VSTA20SDK.exe
16:26 diakopter okay, so, want to hear more details of the runtime design?
16:29 diakopter it would help me to write out the ideas to you|someone
16:30 diakopter so anyway, I'll write them out anyway; feel free to reply/comment anytime
16:40 diakopter my overall design philosophy is based on the principle of absolutely MINIMIZING any upfront "performance" optimization.  That is, optimizing for minimum runtime performance optimization.  It doesn't matter if there's 8 layers of indirection/abstraction, 500 method calls to call 1 real method... I shouldn't care.  As long as each layer is able to have itself reimplemented in a deeper layer, it can be horribly non-performant initially, if making all the
16:41 * diakopter reboots; brb
16:54 diakopter (back)
17:01 diakopter anyway,
17:02 diakopter <sigh> I think I'll just work on coding it directly since I'd end up doing literate programming into IRC anyway just to explain it
17:05 cj yeah, I like hearing details about the runtime design :)
17:06 cj sorry I've been unresponsive.  trying to catch up on work here :)
17:09 diakopter :)
17:35 cj the ball is rolling around ms with your question.  sounds like the free-to-redist bits are the runtime, not the sdk/designer bits
17:36 diakopter erm wow
17:37 diakopter didn't realize you'd spawn an email thread for it.. :P
17:38 diakopter confused
17:38 cj hmm?
17:39 diakopter I'll have to look at what's in the vsta runtime.
17:42 cj He can talk to Summit Software.  Their website is www.summsoft.com and a form to get more licensing information is available at http://www2.summsoft.com/vstainfreq.asp
17:42 cj sokay?
17:46 diakopter yeah I did all that
17:47 diakopter created an account, submitted a product evaluation inquiery
17:47 diakopter inquiry
17:47 diakopter :D
17:48 diakopter so maybe I'll get an email or phone call or something :)
17:48 diakopter ah well
17:49 diakopter I'm just trying to cover all my bases by looking at everything that's available.  I'd really hate to be duplicating effort, esp if the fruits of that effort is freely available/redistributable.
17:57 diakopter and if the vsta runtime .dlls are fully managed (and run on mono), even better ;)
18:00 diakopter I guess my confusion is that it says a 3% royalty must be paid for product sales between $0 and $xxxx..... so if my revenue from my VSTA-linked product is $0 (freely available), wouldn't that be 3% of $0?
18:00 diakopter I guess I could post a question on their forum.
18:04 diakopter I mean, I realize a separate agreement would need to be in place
18:05 diakopter but does that mean they wouldn't agree to someone using their software to make a free product?  seems odd.
18:07 diakopter lol; the EULA says it can't be used to make software to run on non-Microsoft runtimes (presumably including mono).  that's unenforceable.  I mean, who's to say what someone would do with the .dll I distribute?
18:08 diakopter feel free to forward those questions on.... :P
18:09 diakopter okay, the "runtime" is vstar20.exe (836KB)...
18:09 diakopter the IDE/SDK itself, which is the base that customers build on to redistribute, is 39MB (vsta20.exe)
18:21 cj whee.  lot to catch up on :) moment
18:22 cj I think they'd prefer that you talk with Summit instead of them.  I don't feel comfortable asking for more from them at this point...
18:22 diakopter k, cool.
18:22 diakopter thanks for the help!
18:22 cj sure :)
18:29 diakopter erm; lol
18:29 diakopter "
18:29 diakopter "If Perl would get us in trouble, we would have been out of jobs long ago. We use it quite a bit, actually, most of our harness is in Perl.
18:29 diakopter "
18:29 diakopter um
18:29 diakopter guess who said that
18:30 diakopter the C# compiler test team
18:30 diakopter at msft
18:31 diakopter I wonder if that's still the case (5 years later)
18:32 cj of course
18:32 diakopter that's awesome.
18:33 cj yeah.  perl is big here.
18:33 diakopter hm
18:34 diakopter and then there's this: http://blogs.msdn.com/shamit/a​rchive/2005/02/01/365217.aspx
19:46 diakopter static void Main(string[] args) {
19:46 diakopter int final_value;
19:46 diakopter for (int x = 0; x < 5; x++ ) {
19:46 diakopter final_value = x;
19:46 diakopter }
19:46 diakopter Console.WriteLine(final_value);
19:46 diakopter }
19:46 diakopter cj: why doesn't the above compile?
19:46 diakopter ;)
20:47 cj hmm?
20:47 cj give me a sec...
20:47 diakopter heh; don't ask anyone... :P
20:47 cj ha
20:47 cj Main isn't public?
20:47 diakopter nope
20:48 diakopter feel free to try to compile it
20:48 cj doing so now :)
20:48 cj not in a class, if that matters
20:48 diakopter well yeah, assume it's in a class
20:48 diakopter and it's the entry point, etc.
20:49 cj final_value has no value
20:50 cj but it would...
20:50 cj strange
20:51 diakopter yeah; the compiler is being lazy about determining 'definite assignment'
20:51 cj because the type system doesn't recognize that the for loop always gets executed 5 times...
20:51 cj is this a bug?
20:51 cj do you think?
20:51 diakopter yeah; it should know that for loops always get executed at least once.
20:52 cj okay.  you want to file it or should I?
20:52 diakopter :P
20:52 cj I need to get a commit in so I can get props on the next release :)
20:52 diakopter I wouldn't know where to file it?  somewheres on msconnect?
20:52 cj oh, csc doesn't do it either?
20:52 cj I was testing with gmcs
20:54 diakopter oh, no.  well, perhaps something about the for loop initialization could possibly catch a resumable exception... or maybe that's how it's implemented... dunno.
20:55 diakopter (no, csc doesn't catch it either).  also, perhaps it's possible to make a for loop that doesn't always execute at least once.
20:55 cj marek seems to think the error is correct
20:56 diakopter okay... probably because one of the two reasons I said above
23:21 diakopter cj: quite the firestorm on #perl6 today...

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