[log in to unmask] wrote:

[bunch of stuff deleted....]

>I just had to try this, and would be really interested if someone could
>explain why the CI changes behaviour whith the change of the HPPROMPT
>variable, if what is shown below is related to John's original question,
>and, if this is to be considered a bug or a feature (:-):

[test cases deleted...]

This is a feature (and a useful one at that) of the CI.

Instead of responding directly to your example, I will give you a
different one (that helps explain the reason for the CI working the
way it does.

For some historical perspective, consider that in the old days there
was no redo stack (or command history) stack, the "REDO" command was
nowhere near as poserful as today, and there was no "DO" command. Also
users were very familiar with the way their terminals worked and could
use the arrow keys, memory lock, local and remote mode, etc
profeciently.

Customers would frequently page back in their terminal memory to find
a command they wanted to execute again (no redo stack), then they
would move the cursor to that command, go into local mode, delete the
prompt, go back into remote mode and hit enter (transmitting the
command for re-execution).

The step of having to delete the prompt required a bunch of extra
keystrokes and was cumbersome (to say the least). So the MPE/iX CI
folks said "Hey that's kind of dumb! Why not just ignore the prompt
and parse the command?" So they did. Since they also added
configurable prompts (by setting HPPROMPT to whatever you want), they
had to make the CI "smart enough" to ignore whatever the prompt is.

If you change the prompt to "ignore me> " and then do the following:

ignore me >echo please
please
ignore me >ignore me> please
please

The prompt will be "eaten" before parsing the command and the command
will execute without an error (assuming it is a valid command).