RE: EXTPGM and performance

["Joe Pluta" <joepluta@xxxxxxxxxxxxxxxxx>]

Thanks Jon.  This is good to know.

Keith, the reason I want to go this route is to create a common
prototype for a client/server infrastructure.  It's actually working out
quite well.  The idea of all servers using a common parameter list is
making the design of the router incredibly easy.  Right now, my router
is 60 lines of code, and I can extend it pretty much indefinitely.

Joe

> From: Jon Paris
> 
> Using protos or simply CALL/PARM makes no difference.  They are both
> resolved out to the same code.
> 
> This underlying method used was changed way back in the RPG/400 days -
> somewhere around V2R2 if memory serves.
> 
> The way it works now (assuming they didn't break the RPG IV compiler!)
is
> that when a variable is used the content of the variable is compared
with
> the value that was used to resolve the current pointer.  If the
variable
> hasn't changed the pointer is used as-is.  If it has changed ....
> 
> So it is slightly slower to use a variable name, since there is an
extra
> compare involved, but unless the value changes that is the only
> difference.