Hi David,
my view regarding shared variables is that they are a creature of
their
time, kind of Dinosaurs.
I have implemented them only for compatibility with IBM APL2 to
ease
portability existing APL programs. You can have a look at testcases/AP*.tc
to see how to use them.
Shared variables are OK for experimenting (and I will fix the
faults that you
will find) but I would not use them for new programs. Native
functions are
more state-of-the-art and easier to use than shared variables.
/// Jürgen
On 06/28/2014 07:14 PM, David B. Lamkins wrote:
On Sat, 2014-06-28 at 14:13 +0200, Juergen Sauermann wrote:
I also doubt that anybody is using shared variables at all because I
haven't received any trouble reports on them.
/// Jürgen
That's true in my case.
Earlier this week I did make a first attempt to understand how SVs work.
In the short term I'll need to construct a workflow that lets me do more
than trivial experiments with SVs. I rely on Elias' gnu-apl-mode for APL
input. To work with SVs I'll need to patch gnu-apl-mode to provide for
additional command-line options (so I can pass --SV --id # to GNU APL).
Or something... It seems inelegant to have to customize each running
instance of Emacs to set up an id.
At this point, I don't have any definite plans for SVs beyond
experimentation to help me understand the strengths and weaknesses of
the approach.