<bhurt>
Well, a monitor is monitoring multiple different microthreads for exceptions. If the first microthread to throw an exception cancels all the microthreads under the monitor, then there is only one exception the monitor will ever catch- but this isn't required. So there can be several exceptions a monitor can catch.
<bhurt>
ita: don't get me started on Sam.
<ita>
bhurt: you do not like lisp ?
* rfischer
pokes the flamebait.
<ita>
hey, that's just an url, it does not reflect my own opinion
<bhurt>
Chia: you need some sort of way to monitor your monitors, so that you can see what exceptions they've caught.
<bhurt>
ita: I work with Sam. I've seen his code.
<rfischer>
Yeah, I see that. Obnoxious.
<ita>
bhurt: oh, so you work with that guy ?
<rfischer>
Ita: Do you have something constructive to say about the post? Thoughts on it?
<bhurt>
Short answer: I don't hate Lisp, but Ocaml is not Lisp (neither is it Java). And if you insist on trying to program it like Lisp, you're going to run into a world of pain. Which is no different from any other programming language which is not lisp.
<ita>
rfischer: yes, his saying python and ruby are nice languages puzzles me
<bhurt>
He says that because he's not using them.
<ita>
i guess, yes, if he really did he would not write that
Torment has joined #ocaml
<rfischer>
The stuff on that list is pretty minor, coming from someone who is coding a lot of Java/Groovy and just wrote a Ruby library...
<rfischer>
The fact that the type system doesn't catch ALL of your bugs isn't a reason the language sucks.
<ita>
well, caml could have contracts
<rfischer>
And it could have better unit test support.
<ita>
maybe someone will wake up one day and add it to camlp6
<rfischer>
And it could have a lot of stuff.
<ita>
bah, unit test support
<ita>
compared to contracts ~
<ita>
i would take the contracts first
<bhurt>
The real complaint is more like: the type system disallows certain lisp colloquialisms that he's used to using.
<ita>
unit tests are overrated
<bhurt>
Now *that* is flamebait.
<rfischer>
Aw, crap.
<ita>
bhurt: explain
<rfischer>
I've got some kind of assembler error.
<bhurt>
Bwuh?
<bhurt>
Send me the code.
* rfischer
kicks around 3.09.2
<ita>
unit tests gives a false sense of security and encourages programmers to code by interpolation, also unit tests must be maintained; contracts make the specifications part of the code which is quite an improvement already
<rfischer>
There's a place for unit tests which contracts don't offer -- contracts are still at the abstract API level, whereas unit tests force you to actually *write* your code.
<bhurt>
Note: Sam complains that type checking gives a false sense of security too.
<rfischer>
I do agree that there's a Cult of the Unit Test right now, and they are overrated: for instance, the claim that you don't need a type system because you have unit tests is just silly.
<bhurt>
Going back to coding...
<flux>
anything but a mathematical proof of the specification and the implementation give a false sense of security?-)
<rfischer>
BHurt: I pitched you the e-mail.
<rfischer>
It's whacky.
<flux>
I wonder if you can proove that a specification fullfills a set of requirements..
<flux>
or if the requirements make any sense in the first place.
<rfischer>
Sooner or later, you're going to hit the halting problem. :-P
<rfischer>
Or the business reality.
<flux>
I don't think so (about the halting problem)
<rfischer>
Either way, your code is going to go a little bit crazy.
<flux>
and then there's the one thing you know, but can't prove: the customer is always wrong ;)
<rfischer>
Except that they're cutting the checks.
<rfischer>
Since they have the gold, they get to make the rules. :-D
<rfischer>
But, y'know, I'm a consultant.
<rfischer>
So that's my life.
<bhurt>
Chia: what architecture are you on? Those look like PPC instructions.
<flux>
so you need to be creative and figure out what the client _really_ wants
<flux>
hm, not wants, but needs
<rfischer>
I'm on OS-X.
<ita>
rfischer: what country ?
<Smerdyakov>
rfischer, it's not like you have "the consultant DNA" and have no other choices!
<rfischer>
Wow. The consultant comment really brought people out!
<rfischer>
I'm in Minnesota, USA.
<ita>
rfischer: that's only in the usa, in europe it is different :-)
<rfischer>
In Europe, the customers don't have the money?
<bhurt>
In soviet russia, them who make the rules have the gold!
<bhurt>
Chia: Intel OS-X, or PPC OS-X?
<ita>
rfischer: no, in europe, making unit tests comes after writing the documentation
<rfischer>
Intel OS-X. Which means the PPC stuff is probably bad.
<bhurt>
I'm having /. flashbacks.
<ita>
(which means, no unit tests at all :-))
<bhurt>
In soviet russia, the documentation writes you!
<ita>
bhurt: and in north corea only elderly people need documentation, yeah we all know /. memes
<bhurt>
You're generating PPC assembly, and I'm willing to bet dollars to donuts that you're on an intel CPU.
<ita>
(and i, for one, welcome our unit test writers on ppc from minnesota)
<rfischer>
Yeah, I am on an Intel CPU. WTF? My OCaml is 3.10 unless I do which, in which case I'm on 3.9.2
<bhurt>
If taking away any sense of security from programmers actually produced better code, then we should install shotguns in the chairs of programmers pointed at their heads, and if they code a bug, bang. Time to call HR, and the janitor...
<ita>
well, sometimes they should get rid of people in the hr too
<rfischer>
Well, it's a lot easier to code when your programming language and framework actively conspire to hide your bugs from/for you.
<ita>
where i work right now the architect has broken the mvc pattern because he has not found the correct use of default actions in validation.xml, so the jsp make database queries (to fill comboboxes)
<bhurt>
My definition of "code" includes "making it work". Hiding my bugs from me simply makes it harder to code.
yminsky has quit []
<ita>
(when static typing annoys you, make it work in xml!)
<bhurt>
Didn't realize Ron was on.
<rfischer>
Ron?
<bhurt>
Yaron Minsky.
henry_r has joined #ocaml
<rfischer>
Hey, I've got some hardcore XSLT/XML Schema action that has officially become "Don't Touch This" at my last contract.
<rfischer>
I wrote it as a throw-away lark to clean up XML logging reports. I should have known better.
<rfischer>
Now it's a standard part of their development process.
<rfischer>
That's Bad Times.
<ita>
xslt, buahaha
<rfischer>
Hey, it's functional!
<ita>
sure, but it is write-only like perl
<rfischer>
True that.
<rfischer>
And it's noisy. Really, insanely, incredibly noisy.
<rfischer>
But it worked for what I was doing: I could tack it into log files and it would render them as nice HTML.
<rfischer>
Only took me 6000 line of code.
yminsky_ has joined #ocaml
<bhurt>
And hello to Ron again.
spx2007 has quit [Read error: 104 (Connection reset by peer)]
foldr has joined #ocaml
<rfischer>
BHurt; Check your chat window.
<bhurt>
Congrats, I think you're a buddy of mine now.
<yminsky_>
bhurt: Greetings.
<bhurt>
Hello.
<bhurt>
BTW, I am who you think I am.
<yminsky_>
bhurt: there was never any doubt.
<bhurt>
It might not have been me- I could have been out of my head...
<Smerdyakov>
bhurt, and I'm who you probably didn't know I am!
<rfischer>
Anyone else here from Jane St?
<rfischer>
(Meaning, anyone other than Brian?)
<Smerdyakov>
yminsky & me
<rfischer>
Do I know you, Smerdyakov?
<bhurt>
So who are you?
<Smerdyakov>
I don't know. Do you? O_o
<Smerdyakov>
bhurt, I am one of the rare IRCers with accurate /whois information.
ygrek has joined #ocaml
Mr_Awesome has quit [Read error: 110 (Connection timed out)]
yminsky_ has quit []
<bhurt>
Online chat peaked with ytalk and has been going downhill ever since.
<bhurt>
Although there was much to be said for phone on VMS.