Re: [Axiom-developer] Sockets, threads and GUI<->Kernel communication

[Scott Morrison]

On 12 Apr 2007 06:41:28 -0500, Gabriel Dos Reis <address@hidden> wrote:

C Y <address@hidden> writes:

| I do agree that threads are a more attractive way to implement the
| client-server setup when dealing with a strictly local copy of Axiom,

If you agree with that, then you must agree with the over-engineering
characterization.  We have more uses of Axiom as local copy than we
have as a truly distributed system over network.  A 101 principle
of engineering is to keep common thing simple.

Mea Culpa.

As the designer and developer of the overly complex socket communication system, I must take responsibility for this mess.  This system was originally implemented on IBM RT workstations running AIX in 1989.  In AIX there were no lightweight threads, so that really was not an option.  But the real motivation for using sockets was that we envisioned running HyperDoc  and Axiom on different machines.  In those days we had powerful mainframes and wimpy workstations.  Of course it was never really implemented to work that way, but the overly complex socket approach was never removed.

There also would have been no HyperDoc if we'd done this a few years later.  At the time web browsers were non-existent.  I'd be in favor of dumping HyperDoc and implementing a browser interface, but I'm not going to do that, and I don't see it happening soon.

I also did the socket interface to the graphics system for the same reason: we wanted to run things of different systems, and again it never happened.

So that's the historical perspective.

I apologize for leaving you guys with this overly complex mess.

-- Scott Morrison