OSDev.org

    if(NeedToFixBug() && IsWeekend())
        GetBored();

God wrote: The world's best optimizing Brainf**k compiler. The compiler is built for all BF enthusiastics, who want to write small, fast and elegant programs in BF. In the future, I'd even extend the language, to at least support File I/O, and probably whatever you suggest.

• You can improve my code quality (which is poor), by donating some code.
• You can improve the over all product (which is just my weekend timepass) by giving out some ideas.

Solar wrote:Do you really want to bloat the language by 25% or more just to add file I/O?

• Prepare file name, such that the current 'index' points to a null terminated "XYZ.ZYX".
• Do a ^

Solar wrote:Well, BF already allows reading from stdin and writing to stdout. Combined with pipes and output redirection, isn't that all the file I/O you need?

Solar wrote:As for usefulness, I've used BF a couple of times as an introduction to elemental programming. No, seriously. At least for the first lesson.

You get to explain "memory as an array", pointers, the difference between incrementing a pointer and incrementing the value it's pointing to, and the equivalence of ASCII characters and certain numbers. You can also show the importance of breaking down a problem into individual steps. All that with a language you can explain with a single slide projected to the wall.

You also show that programming isn't all about being serious. And you get your audience craving for a higher-level language, so when I toss C++ at them the next lesson, they're actually thankful.

Shikhin wrote:/me suddenly starts hoping that Solar would introduce BFK in his next "first lesson".

davidv1992 wrote:For example combinations like the following are often found in BF programs: [->>+>+<<<].

davidv1992 wrote:Optimizing these into moves can make quite a difference in performance. You could even try to start recognizing more complex patterns such as additions and maybe even multiplication.

Solar wrote:Oh, another message that BF helps driving home: comment your source.

Shikhin wrote:

davidv1992 wrote:[->>+>+<<<]

Please excuse my ignorance, but what would you optimize that to?

Solar wrote:Copying the value of address 0 to address 1 and 2...!?!

Solar wrote:

Shikhin wrote:

davidv1992 wrote:[->>+>+<<<]

Please excuse my ignorance, but what would you optimize that to?

Copying the value of address 0 to address 1 and 2...!?!

d[2] += d[0];
d[3] += d[0];
d[0] = 0;

XenOS wrote:Wait a second... Are you sure? o.O I just tried to understand this tiny bit of code and I think it rather does something like this (where d is the initial data pointer):

Code: Select all

d[2] += d[0];
d[3] += d[0];
d[0] = 0;

d[2] = d[0];
d[3] = d[0];
d[0] = 0;

davidv1992 wrote:Of course, in case the compiler can prove that d[2] and d[3] are zero upon entering this piece of code it can be simplified to the version with just assignments, giving the advantage of saving two reads from memory.