Assert in mtproto-utils.c line 101
Hello,
I encountered the assert(0) in mtproto-utils.c line 101. It says "As long as nobody ever uses this code, assume it is broken."
I had to remove the assert and it worked fine. I guess it's ok to remove it?
Interesting. The original code there was horribly broken (instant segfault). The current version is from me, in the hope of being somewhat okay. However, I don't know of any platform that has 32-bit long but 64-bit long long, so I assumed that this check was pointless anyway.
Could you please explain what's going on with your platform? (I'll post some code to determined your sizeofs in a moment.)
Code to test it (you may need to write llu instead of lu):
#include <stdio.h>
int main() {
#define SAY(x) printf("sizeof(" #x ") = %lu\n", sizeof(x))
SAY(int);
SAY(long);
SAY(long long);
SAY(unsigned int);
SAY(unsigned long);
SAY(unsigned long long);
return 0;
}
Here's the output on AMD64:
$ gcc -Wall -Wextra -pedantic -O2 -g -o say_sizeof say_sizeof.c
$ ./say_sizeof
sizeof(int) = 4
sizeof(long) = 8
sizeof(long long) = 8
sizeof(unsigned int) = 4
sizeof(unsigned long) = 8
sizeof(unsigned long long) = 8
$
What platform are you using? What does it say there?
Btw: You may also want to delete the comment Here be dragons (no longer untested, if you say it works) and the comments in ull2BN
I'm on i686, Ubuntu 14.04 on virtualbox. My output is like this:
sizeof(int) = 4
sizeof(long) = 4
sizeof(long long) = 8
sizeof(unsigned int) = 4
sizeof(unsigned long) = 4
sizeof(unsigned long long) = 8
I removed the other code too, works fine for me. Maybe it's because of virtualbox or something, it is indeed weird.
Not weird, I just thought that i686 worked differently (specifically I always thought that long was 8 bytes there).
So yeah, I'd like to green-light this PR. However, please note that @vysheng seems very busy; I've been waiting for him to merge #94 and #98 for quite a while.
I thought that too :) Thank you, we'll wait for @vysheng then.