Re:Smallest x86 ELF Hello World
by breadbox on Tuesday October 11, 2011 @03:04PM
(#3099858)
Many years ago I used to have a 58-byte version, but around the time that x86-64 was being developed, Linux started examining (and thus validating) a couple more fields in the ELF header. It's really a shame, because the corner-cutting in the 62-byte version isn't nearly as impressive (in my own opinion).
Re:Smallest x86 ELF Hello World
by sambeau on Tuesday October 11, 2011 @01:21PM
(#3099428)
It makes more sense for VM byte-code as the bytes not only have to be loaded (probably by the page, again) but the bytes have to be run and possibly jitted, too.
Re:Smallest x86 ELF Hello World
by sambeau on Tuesday October 11, 2011 @12:08PM
(#3099129)
I edited out the words "apart from on an embedded device with a tiny memory" as I had hoped that the mention of the modern computer and virtual memory would suffice.
But this was a genuine question rather than a put-down. I was hoping to have my assumption debunked.
EDIT: I have no idea why you're being down-voted seems unfair
Re:Smallest x86 ELF Hello World
by VeXocide on Tuesday October 11, 2011 @12:52PM
(#3099315)
You are right in that
31 c0 xor %eax, %eax
is shorter than
b8 00 00 00 00 mov $0, %eax
but that's not the main reason it's often used in shellcode. The xor reg, reg is preferred because it means the resulting shellcode will not contain a null byte. This is desirable since the null-byte is often used as a string terminator, in environment variables for example.
Re:Smallest x86 ELF Hello World
by ars on Wednesday October 12, 2011 @06:34AM
(#3102358)
Because it doesn't make any sense? What, are you planning just randomly try every permutation of 142 bytes?
You can't have a genetic algorithm run on something where almost every attempt will simply produce invalid code. Genetic algorithms work when you can try various strategies to maximize a fitness function. But here the fitness is yes or no, does it work or not. There is nothing to maximize - and you can't try to maximize the "size fitness" because pretty much all the attempts will produce invalid code, so there is nothing to mix together for various generations.
Re:Smallest x86 ELF Hello World
by joelthelion on Sunday October 16, 2011 @05:21AM
(#3116939)
I disagree completely. The fitness function is clear, and that fact that many children will produce invalid code is not that much of a problem, really. You just eliminate them.
Google "automatic programming" if you think it's never been done, your mind is about to be expanded.
You should really learn to think for a while before you dismiss ideas as "nonsensical".
Re:Smallest x86 ELF Hello World
by yan on Tuesday October 11, 2011 @11:38AM
(#3099010)
I believe you entirely missed the point. This was more of an "Look at how you can abuse the loader and play with ELF headers" and in general show an interesting 'hacky' article, not "this is smaller, thus more efficient" argument.
Re:Smallest x86 ELF Hello World
by garethsprice on Tuesday October 11, 2011 @02:17PM
(#3099689)
I figured this was posted as a related article to the Dash "Hello World" issue where people are pointing at 17,000 lines of Hello World code, because it includes the runtime when compiled to Javascript.
There was another comment about how this will be the next "Fibonacci debacle", related to how Ted Dzubia trolled the node.js crowd by pointing out it's weakness for CPU-intensive tasks, then had a ton of people wasting time on writing faster Fibonacci algorithms in node just to prove him wrong.
My point was that comparing languages by the size of their Hello World programs is a useless exercise.
Re:Smallest x86 ELF Hello World
by delinka on Tuesday October 11, 2011 @11:33AM
(#3098980)
Your implementation is not x86 ELF. Regardless, I don't recall the author's insistence that finding the smallest executable was desirable. "Stupid metrics" make for great educational exercises.
http://www.muppetlabs.com/~breadbox/software/tiny/
(62-byte hello world among other things)
http://www.muppetlabs.com/~breadbox/software/tiny/teensy.htm...
http://www.returninfinity.com/baremetal-helloworld.html
Why use ELF or Linux? Flat binaries (and a custom OS) to the rescue! :)
As with all games (even Nomic) a game is only a game if there are rules that limit what one can do.
Surely, once you get below page-size (e.g. 4K) you are just wasting memory that will get loaded anyway?
Compile "Hello world" into an executable, and see how many kilobytes it takes on most architectures, despite the object file being a few bytes.
This thread pretty much responds to that. It is no 17000 lines, nor 17000K, it is just 142 bytes (other comments suggest it is 62).
[1] http://news.ycombinator.com/item?id=3098660
But this was a genuine question rather than a put-down. I was hoping to have my assumption debunked.
EDIT: I have no idea why you're being down-voted seems unfair
The 8086 assembly tutorial book I used then recommended to use "sub ax, ax" instead because it was less surprising.
You can't have a genetic algorithm run on something where almost every attempt will simply produce invalid code. Genetic algorithms work when you can try various strategies to maximize a fitness function. But here the fitness is yes or no, does it work or not. There is nothing to maximize - and you can't try to maximize the "size fitness" because pretty much all the attempts will produce invalid code, so there is nothing to mix together for various generations.
Google "automatic programming" if you think it's never been done, your mind is about to be expanded.
You should really learn to think for a while before you dismiss ideas as "nonsensical".
https://secure.wikimedia.org/wikipedia/en/wiki/Genetic_progr...
(http://codegolf.stackexchange.com/)
By this stupid metric, PHP is the greatest programming language ever made.
12 byte Hello World must mean that it's also the ideal solution for every other programming problem, right?
There was another comment about how this will be the next "Fibonacci debacle", related to how Ted Dzubia trolled the node.js crowd by pointing out it's weakness for CPU-intensive tasks, then had a ton of people wasting time on writing faster Fibonacci algorithms in node just to prove him wrong.
My point was that comparing languages by the size of their Hello World programs is a useless exercise.
Second, there is no point in using it for that when you have bash ... because bash has echo built-in and bash is a programming language:
$ echo "Hello World!"
1 line bitches! I can feel my unix beard growing ...