425. hajime

[sparr0]

it looke like I am not the only one having trouble with this one. I have gotten my answer down to a single space, but hours poring over C references have not been fruitful beyond this.

[uau]

The problem was only added less than two weeks ago. I think giving hints for this problem would be a bad idea since it's so simple and hints would make it trivial.

[sparr0]

when a problem specifies "C Code" but doesn't use the judge's gcc, does it have to be C for that version of gcc?

[mauke]

No, this is standard C. You can't use any gcc extensions, so something like "typeof" won't work. Btw, both of your solutions are buggy, sparr0.

[sparr0]

Hmmm... What is the authoritative reference for your "Standard C"?

[mauke]

The ultimate "reference" is my judge program.
Both solutions are valid C90 (ISO 9899:1990) and C99 (ISO 9899:1999). At least one of them wouldn't work in certain traditional (pre-standard) compilers but I doubt anyone of you uses a >20 years old compiler.

[abhijit_sof]

i have figured out one solution ...........
but the other one i am finding it tough .........

[sparr0]

the hardest part of this problem is testing. since mauke says both of my solutions are buggy im guessing something i used is part of GCC extensions but not C99, which means ive found 0 instead of 1 solution. if i had a "standard" compiler to throw ideas at id probably take another look at this one, but until then im not quite up to the task of re-reading the standards.

[UncleTimmy]

the hardest part of this problem is testing. since mauke says both of my solutions are buggy im guessing something i used is part of GCC extensions but not C99, which means ive found 0 instead of 1 solution. if i had a "standard" compiler to throw ideas at id probably take another look at this one, but until then im not quite up to the task of re-reading the standards.

-Wall -ansi -pedantic

to gcc does a good job of identifying non-ANSI constructs. Give it a try, anyway.

[uau]

since mauke says both of my solutions are buggy im guessing something i used is part of GCC extensions but not C99

Or maybe it means that your answer doesn't actually declare the right kind of type under any compiler.

[misof]

I found the obvious/**/solution, any hints on the second part? Especially, a clarification what is and what isn't whitespace would be really nice.

[mauke]

Huhu, I found the "obvious" other solution first; the /**/ thing was an afterthought. I'm pretty sure that whitespaceless C is turing-complete, even without preprocessor tricks like comments.

"Whitespace" is a sequence of 1 or more white characters. White characters recognized by the judge include: ' ' (space), '\t', (tab), '\n' (newline), '\r' (carriage return), '\f' (form feed) and '\v' (vertical tab). No other ASCII char is white, but there may be other white characters such as char 160 (non-breaking space); I haven't tested those.

A general comment: '\0' (NUL) isn't whitespace. It can't be used to separate tokens either. (I don't care if this works in gcc.)

[diver_ru_sof]

I have small question. Can I use #define directive as type definition?

[mauke]

No, but I'd like to see you use #define without putting whitespace between it and the following identifier...

Oops, I know how to do it, but it's still not allowed: The problem description asks for a declaration/type definition. #define is just token replacement.

[diver_ru_sof]

Ok. And another one.
Does right solution work correctly on C++? (I'm just have no normal C compiler for experiments:) )

[ar2rd]

I have a question.
How much different should the two submissions be? My submissions gives WA even if I used two distinct tricks.

[nneonneo]

Pardon the revival of this thread, but do we use
"char * const"
or
"const char *"
? The description is rather ambiguous about this, and I think I am being rejected because I may have the wrong one (it makes a huge difference to the token-parsing and stuff, and the amount of hackish things you need to make it work)

[darksoul_sof]

nneonneo, it's char*const. meaning a pointer to char that can't point anywhere else, but the thing pointed to can change.

besides /**/, what else have you come up with?

[nilayvaish]

I came up with a solution ( and read misof's solution for the other trick). I am able to compile the solution with gcc 3.3.3 but I am getting WA.

[saurabhcc]

typedef/**/struct{unsigned/**/aku;char/**/*const/**/soku}zan;
typedef/**/struct{unsigned(aku);char*const(soku)}zan;

Can someone point out whats wrong with these declarations....I am going mad with this problem.... spoj.pl/status/HAJIME,saurabh8c/