Hello Lukasz,
On Sat, 22 Nov 2014 07:56:35 +0100, Lukasz Majewski l.majewski@majess.pl wrote:
Agreed in general, but not for this one, since "fixing" is the carpet,
I assume that you are presenting below an answer to a "general" case.
However, as Thomas pointed out earlier, this "fix" is perfectly safe regarding the underlying kwbimage code.
Jeroen and I (full disclaimer: we have discussed the topic on IRC) do not contend that the proposed fix would be unsafe; it *is* safe, i.e. it does not adversely affect the code behavior in any measurable way.
What we contend is that the fix be the /right/ fix (although Jeroen and I have slightly differing criteria for defining what "the right fix" would be).
and the only justification I see as acceptable for doing so is when leaving the warning enabled would cause an obnoxiously high number of false positives.
Well let me add, if "fixing the warning" causes real error to be hidden, we shouldn't "fix" the warnings by modifying valid code.
Each subsequent "fix" for this kind of warning should be considered case by case IMHO, therefore I agree with Albert.
Jeroen also agreed on IRC that disabling the compiler warning is not the right fix either; and I agreed that there had to be a better fix than pseudo-initializing headersz. I therefore suggested refactoring kwbimage_set_header in order to ensure gcc does not emit the warning, but without resorting to non-functional code such as a functionally meaningless initialization.
Problem is, to refactor the code, one needs a gcc which emits the warnig. I tried various versions of gcc (4.7.4, 4.8.3, 4.9.1) and all remained silent when compiling tools/kwbimage.c.
Hence my request: Lukasz, which toolchain are you using exactly? Where can we download it from?
Regards, Jeroen
Best regards, Lukasz Majewski
Amicalement,