> I never saw a prompt that ended in an overstrike character, although
> the bare quote-quad prompt consists of starting a new line without
> typing anything at all. We could find out what the last character
> should be in any other prompt, and we could begin to reply as soon as
> we saw it.
>
> As I understand the discussion, you have been asking for a technical
> solution to a problem that seems to me much easier to deal with by
> convention. That is, if nobody insists on such bad design as
> minimally distinguishable prompts, users will never be confronted
> with the problem, and we don't have to solve it.
>
This might be a nonproblem if a prompt that ends in an accented or otherwise
composed character is "highly unlikely" in real life, but in my experience
the close-your-eyes-and-hope-for-the-best approach is rarely good policy in
a field like telecommunications, where Murphy's Law rules. I'd love to
insert a good example here, but the fact that one does not spring to mind
is, perhaps, a good illustration of how telecommunications systems are
designed and built on the limited experience of (often) an English-speaking
engineer without consideration for how they might be used in other locales.
But again, regardless of the likelihood that a prompt will end with a
composed character, if the script execution engine is based on Unicode,
it must either violate Unicode rules (by ignoring the POSSIBILITY that a
combining character will follow) or else risk false positives and deadlocks.
- Frank
This archive was generated by hypermail 2.1.2 : Tue Jul 10 2001 - 17:20:53 EDT