ti, 2005-02-15 kello 10:19 +0100, Marco d'Itri kirjoitti: > On Feb 15, Justin Pryzby <[EMAIL PROTECTED]> wrote: > > > I suppose I will start filing minor bugs against packages that do > > this. I'd like to hear other people's opinions, though. (It occurs > > to me that help output to stderr is arguably appropriate if an invalid > > option is given). Part of the problem is that its fairly depressing > WTF? This is a long-time UNIX tradition, I'd summarily close such a bug > opened on one of my packages.
In my opinion, if you give a wrong option (or do some other syntax error on the command line), the proper thing to do is to give an error message, preferably short, saying what the error was and how to get the full help text. This is an error message, so it should go to the standard error output. When the user explicitly requests for a help text, it is not an error and should go to the standard output. GNU tools work like this already, and have, to the best of my knowledge, worked like this for well over a decade. Tools that behave differently are also fairly common, so I guess tradition isn't clearcut here. I do claim that the GNU way is the right way here. -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]