[avr-chat] Re: [avr-libc-dev] [bug #32322] snprintf() Hex conversion field width specifiers malfunction

[Rick Mann]

On Feb 1, 2011, at 16:02:06, Frédéric Nadeau wrote:

> It is my understanding that the reported behavior is normal.
> 
> using gcc 4.5.0 on OpenSUSE 11.3, the following gives the same result
> #include <stdio.h>
> 
> int main(int argc, const char** argv) {
>   printf("%#02x\n", 0);          //  Writes "00"
>   printf("%#02x\n", 1);          //  Writes "0x1"
>   printf("%#02x\n", 0xb1);       //  Writes "0xb1"
>   return 0;
> }
> gcc -Wall main.c
> ./a.out yeld:
> 00
> 0x1
> 0xb1
> 
> which is the same that you have provided in the code's comments.
> - "%#02x\n", 0 - the '#' only works for non-zero value, so it prints
> the value 0 with a leadind 0.
> As for the other two, remember than the "0x" is part of the width. If
> you actully want 0x01 use:
> %#04x\n", 1

Ah! interesting. Fair enough, regarding the field width. In the past, I've 
always written "0x%02x", and only started using the "#" last night as I was 
logging my parsing issues.

I actually don't need the "#" flag for my purposes. The problem I initially 
encountered was that "%02x" didn't zero-pad. That might've been due to the 
minimal library not doing padding, although I'd argue that's probably 
reasonable functionality (more so that "#" or "*", for example) for a minimal 
library.