Threads? [was] Re: displaying messages returned from "shell" and Re: Socket Time Out Interval

[Dar Scott]

On Friday, January 30, 2004, at 02:32 PM, Alex Rice wrote:

> On Jan 30, 2004, at 2:22 PM, Scott Rossi wrote:
>> But overall, using "send in" and "wait with messages" constructs
>> allow the equivalent of threaded events.
>
> It's good for things that are coded in transcript, but it does not 
> work if you are dealing with blocking IO, shell scripts, system calls, 
> etc. - things that need to wait, but are not coded in transcript.

Yes.  For example, opening the internal modem on OX X locks for 3 
seconds.  I believe this even occurs on a non-blocking open for some 
reason.  I might be doing something wrong.  If so, I seem to be in good 
company.

However, I think it reasonable to keep Revolution to an event model.  
(Though I might be convinced otherwise.)

Threads might be pushed beyond these two barriers:

Enhanced 'open process' to command-line Revolution might be a way to do 
this.  This is begging the question in Barry's case, because that is 
where the weakness lies.  But, with this, many doors will be opened.  
(By "enhanced" I mean empty buffer vs closed pipe issues are resolved 
and 'open process' separates input and output pipes for Transcript 
handling.)

The other place is in threads created by and managed by externals (and 
typically written in C).  I have put that off in my projects, and want 
to resolve some things first, but this might end up being a way to go.

Dar Scott