hacker.org

Mmh, I wonder how call stack overflow proof the VM used to validate the solutions really is...
My program surely will eat up call stack space like mad:

00^<2?2gp!1+1c

I wanted to post my code aswell, then realized I hadn't saved it. So I go open the challenge and start typing code, my solution turned out to still be in the browser history buffer.

00^<4?1+1cp!

Yes, mine abuses calls aswell.

i dont think the "!" is necessary, not if its at the end of the program anyways

Looks like I have to find out how the call stack trick works
Til then the "straight forward" version that cleans up the stack:

00<0:26*?1+0^044*3+-gp

My first version subtracted 1 because strlen in C is defined as length
without the trailing \0! Maybe this should be pointed out in the challenge...
it's quite confusing to get the "right" answer rejected.

No, it does not count \0.

And you don't need to subtract 1 from the \0 position.

If the given string is "abc" for example, the \0 pisition is 3, and the length of the string is also 3.

mem[0] = 'a'
mem[1] = 'b'
mem[2] = 'c'
mem[3] = 0

Ooops. You're right... another Obiwan (off by one) error.
Somehow I'm attracted by those things.

I finally made it right.. even with slightly wrong thoughts.

Thanks for pointing this out!

I think writing programs for the Hack VM even worse than writing assembly. What's the point of this programming language?

However, what do you mean by 'call stack overflow' and 'abusing calls'?

By the way, this is my program:
http://img3.imagebanana.com/img/o2fuqhh ... nshot1.png

The obtuseness bugged me, so I wrote a quick assembler for it; to help out with HVM problems.
It let me write strlen in 10 seconds! [wrote the assembler originally for the King Mouse problem].

Now I can write code like:
and let the assembler do the job of translating that into for me. I need to make the linker a bit smarter so rather than reserving lots of space for a link constant and then just filling what it needs, it uses a smarter algorithm to calculate inner constants first, but oh well, this works . Its never going to beat the size limit problems, but it makes the no-size-limit-ones much more fun.

I tried writing an assembler that also had a few more utility functions (!= loops, for example). It was meh... It was easier to do them by hand.

(Reminder that adum doesn't want us sharing assemblers, but talking about them, I presume, is fine)

Now whats the matter with abusing call stack ?
My program needs exactly the same cycle-count but seems much bigger
However its kinda bad when the string is too long... storing some counter at 729 and pointer at 730

The thing about the call stack is simple: "g" and "?" are relative jumps, but "c" is an absolute jump that places a return address onto the call stack. But there is no law to use "$" whenever you used "c". And if you don't return from your subroutines, the stack fills up.
Since all the challenges I saw do not require to cleanup anything (they only need to terminate without error), "abusing" the call stack is a legitimate way to achieve certain goals.

Regarding the assembler: There are plenty of HVM challenges, and if I was able to write an assembler/compiler for HVM I would do that (King Mouse is still on my todo list). Fortunately there is the Java IDE which makes it a lot easier to write and debug HVM programs.

I don't think that adum cares about how the challenges are solved (as long as you do it on your own) ... you know... this is hacker.org. But maybe I'm wrong... adum?

I know what a callstack is and how it works

What im wondering is why bother using it, when using variables in memory is executed at the same speed and much less complex.

E.g. to make your code faster or shorter... watch out for the next HVM challenges

My version don't use the call stack.

00^<8?1+058+-gp

=/