On Wed, 2002-07-17 at 04:35, Anton J Aylward, CISSP wrote:
Actually, at least in my version (1.1.0.99), they move to the prev/next message, or move to the prev/next folder, or scroll the current message up/down, etc., depending on where the focus is. I think it would be better if there were separate keybindings for prev/next read/unread message, and prev/next read/unread folder.=20 Possibly up/down arrow should always scroll the current message. It's too much of a pain to shift focus for the up/down arrows to have so many roles.Please no! This goes against all the accepted rules of user interface design, man-machine interfacing, ergonomics or whatever you want to call it, not least of the principles based on actual field research.
I hate the way people are beaten over the head with "accepted rules of
user interface design". It is usually used as a way of dismissing
suggestions without considering them in detail. (I'm not necessarily
talking about your reply in particular.) For instance, in this
situation, instead of asking what would happen if "N" moved to the next
message, instead of wondering if new users would be confused by this,
and instead of considering the convenience to users familiar with the
mailer, we immediately move into dubious metaphors...
Meanwhile to go to the next message I have to press Alt-A, G, N.
Somehow, as I type four times as much as I should, I cannot find much
consolation in the fact that Evolution adheres to "accepted rules of
design". Is one of these "accepted rules of design" that the
application has to be a pain to actually use?
The "context" you describe has been found to be the way people really work. The meaning of "shoot" depends on whether you are holding a gun, holding a camera, or playing soccer.
It's interesting that you don't mention that languages have tens of
thousands of words, and not five that each mean different things
depending on context. And why did you pick "shoot" instead of
"computer", "keyboard", or "desk" (just to name three things directly in
front of me)?
But the main problem with the analogy is that in reality, there are
not different contexts when the focus is on the folder versus when it is
on the message scan window. If you ask someone playing soccer or
holding a camera what they are doing, they (after being confused maybe)
will probably say that they are playing soccer or taking pictures. But
I imagine no one thinks of themselves as "being in the folder selection
widget" as opposed to reading a message.
Someone (Rob Pike perhaps?) described the kind of keybinding Ben is talking about as having a dashboard on your car with hundreds of buttons labeled things like "turn on radio and search for some good music; if not found then pull over at the next drive-through donut store".
There are again tons of problems with this analogy. Firstly, it's not clear what driving and reading email have in common. Secondly, the number of buttons on a keyboard is fixed, I was just proposing that some of them do something useful. Thirdly, I wasn't proposing that one key do two separating things like "turn on the radio and ... pull over" I was suggesting that one key do something like "next folder" or "next message" - is that so complicated? Fourthly, car radios have almost all the bindings I describe: AM/FM=Change Folder, Tuning Wheel=next/prev message, Seek=next unread message. In fact, many radios have station presets, a feature that would be really useful in evolution.
If you really want all those keybindings, why don't you use EMACS as your MUA instead of Evolution? EMACS is good at keybinding and complex context-insensitive operations.
This is irrelevant for a number of reasons: it is a category mistake
since emacs isn't a (single) mailer; it is a personal attack of the
"love-it-or-leave-it" variety (which just stifles improvement); and it
ignores the obvious possibility that I like something about Evolution
(vfolders) besides its keybindings.
But it also suggests that emacs has such an execrable interface that
anything that made Evolution resemble Emacs in any way would be a
disaster. But has anyone ever considered that there is *something* good
about Emacs' interface? Or wondered why some many thousands of
sophisticated users use Emacs even though it is apparently such a POS,
and even when they could easily which to a different editor? If it is
so obviously unusable why has it been ported to dozens of platforms? No
one could be productive using Emacs - but then why do thousands of
programmers use Emacs, many of whom care about their productivity and
maybe could even write their own editor?
Of course I'm not suggesting we turn Evolution into an emacs GUI.
But we should recognize that Emacs does some things right.
--
Ben Escoto
Attachment:
signature.asc
Description: This is a digitally signed message part