Hi Bruno, Bruno Haible <[email protected]> writes:
> > c. This will cause compilation errors in two existing packages [1][2]. > Which is undesired for Gnulib modules. Yes, this part especially is not good. > In summary, this patch causes semantic trouble that is hard to fix. > > 2) What kind of patch is this? A bug fix or an optimization? > > As I see it, it is merely an optimization, and in particular one for a > contrived, unrealistic use-case. 99.999999% of the 'date' command use > a small number of format directives, and therefore it is irrelevant > whether they continue the processing after an earlier format directive > failed. > > Based on 1) and 2), I am opposed to this patch. The optimization is not > worth the compilation errors in the existing callers of this function > and the potential bugs caused in the future callers of this function. Thanks for your feedback. I think that it is still better if 'date' exits immediately upon write errors. But the behavior probably doesn't make sense for all callers. I'll think about it some more and see if I can come up with something better. Collin
