Il 05/10/2012 14:41, Luiz Capitulino ha scritto: >>> > > That DPRINTF() usage is really bizarre, it seems its purpose is to >>> > > report >>> > > an error to the user, but that's a debugging call. >>> > > >>> > > I'd let it there and replace it later with proper tracing code, but >>> > > that's >>> > > quite minor for me. Please, at least mention you're dropping it in the >>> > > log. >> > >> > This one is not dropped, it becomes the reported error message. > What I meant is that the error/debug message won't be printed the same way > it was before. This is an improvement, but it's a good idea to mention it. >
I mentioned it for fd, but the popen() function only returns NULL if the fork(2) or pipe(2) calls fail, or if it cannot allocate memory, so I have no idea how to trigger it. >> It would also fix the clobbering of errno. > > I guess so. Is it possible for popen() to return success but then set the > FILE's fd to -1? No. I will add the assert(). Paolo
