Matthieu Moy <[EMAIL PROTECTED]> writes:

> Mark Triggs <[EMAIL PROTECTED]> writes:
>
>> Yep, this sounds like a good idea.  We would still need to deal with the
>> case where the user goes ahead and closes the buffer anyway
>
> I'd like something like
>
> A tla process is running in this buffer. Kill anyway? (y, n, k, ?)
>
> With the binding:
>
> y => just kill the buffer normally
> n => don't do anything
> k => kill both the buffer and the processes running in it
> ? => This help message.
> (and why not another key to open the log buffer)

I like the sound of this (although I would probably swap 'y' and 'k' so
'y' kills the buffer and the process).  Do we have any way of knowing
whether a buffer currently has a process associated with it?  For
example if I start a "tla-missing" and four tla processes get spawned,
is there any way of knowing that those four processes correspond to the
*tla-missing* buffer?

This still doesn't quite stop the problem of the user killing the buffer
but not the corresponding process (although it makes it less likely).
Removing the (with-current-buffer ..) from TLA--RUN-TLA-ASYNC and
relying on process handlers to manage their own buffers still seems to
be the only reliable way to stop these problems.

Mark

-- 
Mark Triggs
<[EMAIL PROTECTED]>


Reply via email to