putnam@thuban.steinmetz (putnam) (07/02/87)
Im more than a little bothered by the keyboard handling in NeWS. The keyboard handling that is already in it is ok for most vanilla stuff and emulates terminal (ie vt100) keyboard handling acceptably (with one exception) - but im not sure that is enough for major windowing applications. The first problem with keyboard handling is the lack of auto-repeat on keys. That is, if i hold down the "f" key, i expect to get "fffffff" until i take it up with some reasonable time between the "f"'s. This does not happen and i have looked at the keyboard handling code and dont see offhand just where to put it. I thought it might be nice to put it in the terminal handling at least for psterm - but that would require even more hackery as there is no provision for handling "up" events. Not impossible, just painful. A more fundamental problem seems to be the handling of shift-control-meta type keys. Currently the shift and control keys are encoded when you get the event (so that "shift-a" becomes "A" and so forth), the meta key is ignored here. It is certainly possible to get the values of the shift, control, meta keys - but is more work (look for shift_keys_down, and so forth in dictionary UI_private). But at this point the key has already been encoded for you and it may be more difficult to get at what you want - that is if you really want to deal with control-shift-meta-left-overshoe-x. Again, its not impossible to get this stuff - but i think the X model is simpler and easier to deal with - though it does require more work on the part of the client to handle. I imagine that one of the reasons for doing things this way is to help to unify the user interface - so that most user interfaces look similar - and this is a laudable goal - but is this the right level at which to do it? Any thoughts? Well, shall we go? -- jefu (jeff putnam) Yes, lets go. -- UUCP: steinmetz!putnam (They do not move.) -- ARPA: putnam@ge-crd.com
rxb@ascway.UUCP (Rafael Bracho) (07/07/87)
Date: 6 Jul 1987 08:28-PST
From: Rafael Bracho <spar!ascway.UUCP!rxb@decwrl.dec.com>
Subject: Re: keyboard handling
To: uunet!steinmetz!thuban!putnam@seismo.css.gov (putnam)
Cc: news-makers@brillig.umd.edu
Jeff,
I agree with you in that auto-repeat is an important feature missing
from NeWS 1.0; mind you, it was present in most of the older beta tapes.
Hopefully Sun will re-instate auto-repeat soon.
As for the shift keys, there is a field in the events called KeyState
which is an array of both keywords and numbers, corresponding to the
raw data obtained from the keyboard. Thus the following event handler
returns a byte encoding the state of the keyboard (recall that the
event is put on the stack by NeWS before calling your handler):
/HandleMouseClick { % event => -
begin
0 % The byte we'll modify
KeyState length 0 ne {
KeyState {
dup /Meta eq {
pop 2#00010000 or
} {
dup /Control eq {
pop 2#00001000 or
} {
/Shift eq { 2#00000100 or } if
} ifelse
} ifelse
} forall
} if
end
} def
(Note that I don't use a case statement because I've found it too slow
for inner-loop applications.)
There is one more problem I've found. Often NeWS misses a shift key
completely, thus you get lowercase when you meant uppercase, or the
string "xs" is inserted in your emacs buffer when you meant to save the
file. This seems to be a synchronization problem, but I haven't had
time to track it down.
Rafael Bracho
RXB@SPAR-20.ARPAian@wcwvax.UUCP (Ian Kemmish) (07/11/87)
From: Ian Kemmish <mcvax!wcwvax!ian@seismo.CSS.GOV> Date: Thu, 9 Jul 87 11:00:23 +0100 To: news-makers@brillig.umd.edu Subject: Re: Keyboard handling Hmmm. We've got NeWS running on machines with cheap IBM clone keyboards on them. (Well, it's one way to keep costs down!) These auto-repeat whether you like it or not, so we definitely don't want the window manager to start repeating keys as well. . . If Sun re-instate auto-repeat keys in NeWS, I hope they do it portably!