Passing negative precision is broken

[Deleted user]

An indirect precision (as in .*) is passed as a signed int. But this code interprets it as unsigned. C11 even specifies that negative precision is taken as if the precision were omitted.

Test case: printf("%.*d", -1, 1) should output 1.

[mpaland]

Good catch!
This case is indeed not implemented. 😞

[ledvinap]

There is still subtle bug: sprintf("%.*d", -1, 0); should produce "0" (precision omitted), not "" (precision specified and zero)

[Deleted user]

Over the time, I found some more subtle issues that (potentially) mismatch with standards or generally expected behavior. They probably don't matter for embedded use at all. Anyway, I used the test suite from this project as base, ported it to C, and added some more test cases, so if you want: https://github.com/wm4/libinsanity/blob/master/tests/printf_test.c
(That project doesn't target embedded use at all.)

[ledvinap]

@wm4 thanks for the link.

I'm now playing with printf, it should be easy to fix and I'll try other cases