On Sun, Aug 20, 2006 at 07:48:29PM +0300, Lars Wirzenius wrote: > su, 2006-08-20 kello 18:08 +0200, Wouter Verhelst kirjoitti: > > On Sun, Aug 20, 2006 at 03:59:17PM +0300, Lars Wirzenius wrote: > > [...] > > > fprintf(stderr, "Could not read file: %s: %s\n", filename, > > > strerror(errno)); > > > > > > Please consider the above a minimal requirement for an error message: it > > > reports the operation that was attempted (reading a file), the target of > > > the operation (the filename), and the error message. > > > > > > perror is, on the whole, useful only in throwaway programs, or programs > > > whose users like guessing. > > > > Bullshit. > > Your version is a common mistake. Calling sprintf may change errno. > perror uses errno, so at the very least you need to save and then > restore errno just before calling perror. Using strerror is just easier.
Still; my point was that it is incorrect to say that "perror is, on the whole, useful only in throwaway programs, or programs whose users like guessing." I gave an example to the contrary, and I think it still holds. As with anything, an API function is not useless merely because most people use it badly. This bad usage should be abolished, not the function as a whole. This is, of course, not the same issue as with things like gets(), which should just not ever be used. But that has a different reason. -- Fun will now commence -- Seven Of Nine, "Ashes to Ashes", stardate 53679.4 -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
