Kari

[turb_]

[QUOTE=Darwin226;25086444]Are there no disadvantages? Is there any benefit from writing a full name?[/QUOTE]

Nope.

[pollyzoid]

Wasn't <= supposed to immediately return? Why doesn't this work then? :saddowns:

>> (str -> if true (<= str) "nope") => test
=> (function)
>> test "yep"
=> nope

[turb_]

[QUOTE=raBBish;25086809]Wasn't <= supposed to immediately return? Why doesn't this work then? :saddowns:

>> (str -> if true (<= str) "nope") => test
=> (function)
>> test "yep"
=> nope

[/QUOTE]

For starters, the return value must precede <=.

Secondly, the <= only returns from the current function, in this case the one passed to if. That's pretty useless, and I'll commit a workaround soon.

[pollyzoid]

[QUOTE=turb_;25086897]For starters, the return value must precede <=.

Secondly, the <= only returns from the current function, in this case the one passed to if. That's pretty useless, and I'll commit a workaround soon.[/QUOTE]

Alright, thanks for explaining.

[editline]04:22PM[/editline]

(b1 -> (b2 ->
	if_else b1 (
		if_else b2 (
			true
		) (
			false
		)
	) (
		false
	)
)) => and

(b1 -> (b2 ->
	if_else b1 (
		true
	) (
		if_else b2 (
			true
		) (
			false
		)
	)
)) => or

put and true true
put and true false
put or true false
put or false false

:v:

[jA_cOp]

[QUOTE=Darwin226;25086444]Are there no disadvantages? Is there any benefit from writing a full name?[/QUOTE]

The "var" keyword invokes type-inference. It means the variable is still fully typed, but the actual type is inferred from the initializer. Hence to use "var", you need to declare & initialize in one statement, after that, the variable has a static type and has no effect on what happens at run-time (unlike the "dynamic" keyword).

I've actually heard a lot of C# programmers complain about the use of "var" because they think it causes inconsistencies and ambiguities in your code.

Just trying to give the whole story here, personally I think that's rubbish and "var" is a god-send in an otherwise quite verbose language.

(Many other languages have type-inference, most notably C++0x has the same behaviour with "auto")

[Darwin226]

So actually using var is nothing more than a shortcut to writing the whole type before the name?

Nifty.

[turb_]

[QUOTE=raBBish;25086933]Alright, thanks for explaining.[/QUOTE]

Did this for you:

[release][b]New syntax[/b]

You can now repeat the '<' in the <= operator to return through multiple functions.

[img]http://ahb.me/x7g[/img]

<= is regular return, <<= returns the function that called the current function, and so on.

You'll need to update your copies to use this feature.
[/release]

It's up on Try Kari too.

[editline]11:55PM[/editline]

I don't know if I should be proud of how I achieved that feature or not...

[cpp]
if (Tokens[i] is ReturnToken)
throw new ReturnFromFunction(((ReturnToken)Tokens[i]).Levels - 1, val ?? NilValue.Nil);

// ...

try
{
val = FunctionStack.Pop().Call(val);
}
catch (ReturnFromFunction r)
{
if (r.ReturnsLeft > 0)
throw new ReturnFromFunction(r.ReturnsLeft - 1, r.Value);

val = r.Value;
}
[/cpp]

[esalaka]

[QUOTE=turb_;25087401]I don't know if I should be proud of how I achieved that feature or not...
[/QUOTE]

Clever use of exceptions there.

[pollyzoid]

I might use this in my C# game engine, with your permission of course. :3:
Since the syntax is really similar to traditional (Source) console commands, it would be useful for a developer console.

[turb_]

[QUOTE=raBBish;25087564]I might use this in my C# game engine, with your permission of course. :3:[/QUOTE]

[url=http://github.com/charliesome/Kari/blob/master/readme.md]I gave my permission here[/url] :p

[ZeekyHBomb]

[QUOTE=turb_;25087401]I don't know if I should be proud of how I achieved that feature or not...

[cpp] if (Tokens[i] is ReturnToken)
throw new ReturnFromFunction(((ReturnToken)Tokens[i]).Levels - 1, val ?? NilValue.Nil);

// ...

try
{
val = FunctionStack.Pop().Call(val);
}
catch (ReturnFromFunction r)
{
if (r.ReturnsLeft > 0)
throw new ReturnFromFunction(r.ReturnsLeft - 1, r.Value);

val = r.Value;
}
[/cpp][/QUOTE]

I don't think you should as that is not what exceptions are intended for. Could you not just return a struct{ bool totalReturn; returnval val; } from the Call-method and check for totalReturn?

[Qombat]

Will this work fully on Mono? I know that it doesn't have quite the ability of the .NET systems.

[turb_]

[QUOTE=Qombat;25106698]Will this work fully on Mono? I know that it doesn't have quite the ability of the .NET systems.[/QUOTE]

Kari is targeted at .NET 4, which is (afaik) currently unsupported by Mono. .NET 4 support will be present in Mono 2.8, and is available in the [url=http://github.com/mono/mono]Mono repository[/url].

I don't believe I'm using any .NET 4 features, so I'll retarget Kari to 3.5 (supported by Mono), run a few tests and if it all works, I'll retarget the whole thing to .NET 3.5.

[Darwin226]

[QUOTE=turb_;25106881]Kari is targeted at .NET 4, which is (afaik) currently unsupported by Mono. .NET 4 support will be present in Mono 2.8, and is available in the [url=http://github.com/mono/mono]Mono repository[/url].

I don't believe I'm using any .NET 4 features, so I'll retarget Kari to 3.5 (supported by Mono), run a few tests and if it all works, I'll retarget the whole thing to .NET 3.5.[/QUOTE]

What are the new features anyways? I think I only used Tuples that were unsupported until 4.0.

[pollyzoid]

[QUOTE=Darwin226;25106931]What are the new features anyways? I think I only used Tuples that were unsupported until 4.0.[/QUOTE]

Dynamic variables, optional parameters and named parameters at least.

[editline]04:49PM[/editline]

Anyways, I just tested and it works perfectly running under Mono 2.6.7 after re-targeting to 3.5 and removing Microsoft.CSharp reference.

[img]http://localhostr.com/files/050083/MonoKari.png[/img]

[esalaka]

[QUOTE=raBBish;25107111]Anyways, I just tested and it works perfectly running under Mono 2.6.7 after re-targeting to 3.5 and removing Microsoft.CSharp reference.
[/QUOTE]

Did you test it on Linux, too? :3:

[turb_]

[QUOTE=raBBish;25107111]Dynamic variables, optional parameters and named parameters at least.[/quote]

The latter two are C# 4 features.

.NET 4 introduced covariance in generics, which I wasn't sure if the marshaling stuff relied on.

[turb_]

So I was writing up a suite of tests for Kari, which included a test similar to this:

[cpp]
[TestMethod]
public void NonObjectAsObject()
{
Test.Throws<RuntimeException>(() => Evaluate(@"1.test"));
}
[/cpp]

I wasn't deliberately trying to break anything, I was just trying to get a few tests up and working.

When I ran the tests, expecting them to all pass, I got this:

[img]http://ahb.me/xQV[/img]

Unit Tests are great :v:

[editline]03:51PM[/editline]

[QUOTE=raBBish;25084484]I tried :v:

(n ->
        put n
        if lt n 2 (n)
        if eq n 2 (add fib sub n 1 fib sub n 2)
        if gt n 2 (add fib sub n 1 fib sub n 2)
) => fib

No idea if it works, it times out on the website and I'm too lazy to test with the real version.[/QUOTE]

Turns out this works (with a minor modification), and I'm using it in a unit test! (I hope that's alright with you)

[cpp]
[TestMethod]
public void Fibonacci()
{
var val = Evaluate(@"
(n ->
put n
if lt n 2 (n <<=)
if eq n 2 (add fib sub n 1 fib sub n 2 <<=)
if gt n 2 (add fib sub n 1 fib sub n 2 <<=)
) => fib

fib 9");

Assert.IsInstanceOfType(val, typeof(NumberValue));
Assert.AreEqual(34, ((NumberValue)val).Number);
}
[/cpp]

[pollyzoid]

What happened to if_else? :ohdear:

[Robber]

Does Try Kari support <<= yet?

[img]http://dl.dropbox.com/u/1106779/TryKariBug.PNG[/img]

[pollyzoid]

[QUOTE=Robber;25127776]Does Try Kari support <<= yet?[/QUOTE]

[img]http://localhostr.com/files/ef44e1/karistuff.png[/img]

Yes, but there's something else going on...

[img]http://localhostr.com/files/e97664/errorkari.png[/img]

[turb_]

Ahh yes, closures were recently broken.

They've been fixed, but I forgot to update Try Kari. Hold on, I'll fix it now.

[editline]10:03PM[/editline]

Fixed:

[img]http://ahb.me/x_y[/img]

[Robber]

I solved the first Project Euler problem :buddy:

(n-> (m-> if m (if n (true <<<=))<=false))=>and
0 => i 0=> sum while (lt i 500)(if eq mod i 3 0 (add sum i => sum) if and eq mod i 5 0 not eq mod i 3 0 (add sum i => sum) add i 1 => i) put sum
while (lt i 1000)(if eq mod i 3 0 (add sum i => sum) if and eq mod i 5 0 not eq mod i 3 0 (add sum i => sum) add i 1 => i) put sum

I had to split the loop in two lines because it always timed out.

[esalaka]

[url]http://downforeveryoneorjustme.com/http://trykari.charliesomerville.com/[/url]

Oh god what did I kill try kari :ohdear:

[editline]04:00PM[/editline]

>> (n->mul 2 n) 2
=> 2

What is going on o_O

[turb_]

[QUOTE=esalaka;25128509]

>> (n->mul 2 n) 2
=> 2

What is going on o_O[/QUOTE]

Can't call a function like that.

If you really want to, you need to dereference it first:

>> (x -> x) => deref
=> (function)

>> deref (n -> mul 2 n) 2
=> 4

It displays two because 2 is the last value evaluated in that expression.

[editline]11:16PM[/editline]

[QUOTE=Robber;25128446]
I had to split the loop in two lines because it always timed out.[/QUOTE]

Grab a copy of Kari Studio or kari_run (the REPL) from the git repo and your life will be so much easier.

[editline]11:19PM[/editline]

Remind me to implement and, or, not and xor in the standard library.

I'll get it done tomorrow.

[pollyzoid]

I'm not quite sure what's happening here... by my logic it should work fine, but you can see the result.

[img]http://localhostr.com/files/e23df3/karistuff3.png[/img]

[esalaka]

[QUOTE=turb_;25128667]Can't call a function like that.[/QUOTE]

I'd ask you to make it possible to call anything evaluating to a function like that, but eh, I can't really see any scenario where it'd be necessary.

[turb_]

[QUOTE=esalaka;25128778]I'd ask you to make it possible to call anything evaluating to a function like that, but eh, I can't really see any scenario where it'd be necessary.[/QUOTE]

It would cause problems with passing them to functions as they'd be evaluated immediately if there was anything following them.

Unless you want to put @ before all your functions, it's probably best the way it is.

[esalaka]

[QUOTE=turb_;25128809]It would cause problems with passing them to functions as they'd be evaluated immediately if there was anything following them.

Unless you want to put @ before all your functions, it's probably best the way it is.[/QUOTE]

Well, is there a function evaluation operator or something?

[turb_]

[QUOTE=raBBish;25128755]I'm not quite sure what's happening here... by my logic it should work fine, but you can see the result.

[img]http://localhostr.com/files/e23df3/karistuff3.png[/img][/quote]

Tested it myself, it's definitely a bug.

I think it's fixed (as the test passes), but not committed to git.

Does the same issue occur in the debug builds of kari_run? They're usually more 'fresh' than the release builds.