hex error

[Keith Thompson]

[...]

Note that any 2-digit string literal would do. If I were going to
use this approach, I'd probably omit the initialization and add
b[2] = '\0';
at the end. (I'd also assign to b[0] before b[1], but that doesn't
really matter.)

b[1] = (unsigned char)c / 16;
b[1] += (b[1] < 10)? '0': 'A';

b[0] = (unsigned char)c % 16;
b[0] += (b[0] < 10)? '0': 'A';

Maybe it's just me, but I'd (also?) put the more significant nibble in
b[0] and the less in b[1]. Or if that's what you meant, I don't think
you were clear about it. (Forest for trees maybe?)

I wasn't thinking about which nibble should go where, just that
assigning to b[1] first and b[0] second seemed odd.

 

[luserXtrog]

[...]

Note that any 2-digit string literal would do.  If I were going to
use this approach, I'd probably omit the initialization and add
    b[2] = '\0';
at the end.  (I'd also assign to b[0] before b[1], but that doesn't
really matter.)
    b[1] = (unsigned char)c / 16;
    b[1] += (b[1] < 10)? '0': 'A';
    b[0] = (unsigned char)c % 16;
    b[0] += (b[0] < 10)? '0': 'A';

Maybe it's just me, but I'd (also?) put the more significant nibble in
b[0] and the less in b[1]. Or if that's what you meant, I don't think
you were clear about it. (Forest for trees maybe?)

I wasn't thinking about which nibble should go where, just that
assigning to b[1] first and b[0] second seemed odd.

Yes. Attempted cleverness bites again.
I expect I'll do something stupid like this again, eventually.
But I will try to incorporate the following lessons.

* Don't do stuff in a funny order.
* Stop assuming ASCII encoding.
* Use library functions to avoid the need for assumptions
about encoding.

I think I was half-way planning for extensibility.
To convert a larger int to a string of hex nibbles,
you'd mod off the the least-significant piece and
then div down the remainder. And then it would be
simpler to fill the string right-to-left.

But the code I posted was indeed garbage.