On Fri, Nov 30, 2001 at 02:02:24PM +0300, Alexander V. Lukyanov wrote: > > I wouldn't recommend that--not for this reason alone, anyway. Note that > > this is exactly the same problem he was having with MKD--he's probably > > receiving a 5xx code here, too. (And we probably shouldn't add > > "ftp:use-mkd".) > > Not for this reason, of course. But it is not a standard command and > I can imagine some server to break because of it.
A server breaking because of an unsupported command? If that actually happened (a server actually dropping the connection due to it), it would be reasonable to have settings for them. But I don't think there have been any reports of that. > > "A rule of thumb in determining if a reply fits into the 4yz or the 5yz > > (Permanent Negative) category is that replies are 4yz if the commands > > can be repeated without any change in command form or in properties of > > the User or Server" > You forgot to mention that RFC states: > > 5yz Permanent Negative Completion reply > > The command was not accepted and the requested action did > not take place. The User-process is discouraged from > repeating the exact request (in the same sequence). Even > some "permanent" error conditions can be corrected, so > the human user may want to direct his User-process to > reinitiate the command sequence by direct action at some > point in the future (e.g., after the spelling has been > changed, or the user has altered his directory status.) > > "User-process is discouraged from repeating the exact request". There is > no ambiguity. But some servers send something like "550 failed to establish > data connection: timeout". That is why I added net:persist-retries. (Incidentally, I disagree: the case where the command can be repeated without any change in command form but doing so is discouraged fits both of these. It's probably best to discard the "rule of thumb" completely. But this is server-stuff; servers interpret the RFC and clients are left to cope with the servers, even if they're wrong.) In any case, it doesn't matter here, as the 550 is being sent in the way lftp expects (in his MKD debug output and the same thing is probably happening with SITE CHMOD for him); everything is working as designed. I think Piotr should just turn off persist-retries unless he has a reason to have it on. I see its use as a workaround, but it's not practical to have on all the time, as it ignores the 5xx definition completely. In other words, I don't think there's a problem. :) -- Glenn Maynard
