Date: Mon, 22 Apr 1996 12:21:36 -0700
From: ron@apple.com (Ron Hochsprung)
Subject: Item #338: Dev Sup Ext: scancode methods


P1275 Openboot Working Group Proposal -- Proposal #:338 Ver <Author-Version#>

Title:          clarification of scancodes and their use

Author:         Ron Hochsprung

Date:           4/22/96

Ed/Tech:        Technical

Synopsis:       make scancode methods usable

Doc & Version:  Device Support Extensions 0.6

Problem:

The definition and use of "scancode" in the keyboard device assume that
a single cell is sufficient to represent a "scancode".  While this may
be true in some systems (i.e., ADB), it is not true in general.

If the scancode is supposed to be the low-level codes that the keyboard
sends, then a single cell is not sufficient.

This "proposal" is primarily put forth to create a discussion item. I'm not
sure what the "correct" answer is.

Proposal:

Define the scancode to be a multi-byte string that represents all of the
low-level codes (dependent upon the keyboard implementation) that represent
one keyboadr "event".  I.e., all of the codes that represent a key-press
or key-release.

If this is the interpretation of scancodes, then change the call to get-scancode
as follows:

    get-scancode     ( adr len msecs -- actual true | false )

where adr,len are a buffer large enough to hold the worst case scancode
sequence.  actual is the length of the scancode sequence returned.

Change the definition of scancode->char to:

    scancode->char  ( adr len -- char true | false )

where adr,len are as above.

As I said, I am primarily introducing the proposal to stimulate dialog, but
I don't think that the current definition "works".

Ron

[ P1275 Item #338 -- Received: Mon Apr 22 12:20:02 PDT 1996 ]