On Thu, 26 Sep 2002 10:09:29 -0700, Vladimir A. Butenko wrote:
> Do you suggest that there should be NO strict and explicit definition for
> the cases when BAD should be returned instead of NO?

My opinion (subject to change with a convincing argument):

It is desirable to distinguish between BAD and NO cases.  BAD means "protocol
violation".  To a great extent, a command which both the client and server
known can not possibly work *is* a protocol violation.

On the other hand, I am not particularly interested in fixing a grey area of
the BAD vs NO choice if there are no obvious client consequences.  Put another
way, issues which cause interoperability problems need to be fixed; purely
academic issues don't necessarily need to be fixed.

> Or - that the NO
> responses should be issued in this and that particular cases, BAD - in this
> and that, and all other error situations may result in BAD or NO, depending
> on the implementation.

This is more or less what is intended, with the vast majority of cases being
clearly NO or BAD and only a small minority being "either".

Reply via email to