. Re: cross-stack globals, also, file inclusion

[Dar Scott]

On Saturday, October 25, 2003, at 04:10 PM, Rob Cozens wrote:

>> To keep this discussion simple, this can represented in XML format.
>
> How would one get values for pointers, handles, and addresses of call 
> back routines via Transcript, Dar?

Ow!

Handles can be simply numbers.  They have no use except to be passed to 
other routines in this scheme.

I guess there should be a description of data going back the other way. 
  That would apply to returned values and to pointers point to values 
that can be modified including pointers.  The description of, say, 
buffers that can be modified can include variables to be modified or 
can include a scheme to pack it up and be part of the returned value.  
This would all apply to the parameters of the callback, too.

If a routine returns the address of a function, that might be handled 
in the same manner as a handle.  Does that really happen?

If callbacks must be unique for calls, you might have to have an 
external call to create a callback handler.  It would be used as 
needed.  It would later be closed.  It could be shared if possible or 
several could be created.  One simple way for it to work would be for 
it to set some global values.  The application could poll those values.

(Another way would be for it to queue a 'send'.  I have figured out a 
way to send, but it is slow.  Also, I don't know if revolution 
callbacks are threadsafe, so these will have to be queued and handled 
by the external idle.  This is all very slow.  See the bugzilla 
enhancement 678.)

Are you going to raise the bar, again?

Dar Scott