From: Hans Aberg (email@example.com)
Date: Sat May 26 2007 - 05:13:44 CDT
On 26 May 2007, at 06:19, Philippe Verdy wrote:
>> Then the next problem is to enter these symbols efficiently.
>> Developing ones own keyboard maps for the large numbers of math
>> symbols takes a lot of time.
>> So I think, best would be some set of Unicode keyboard maps, just
>> designed to enter the characters as efficiently as possible. These
>> might then move away from traditional QWERTY-like keyboard maps for
>> ASCII: ABCDEF seems the right thing, here. :-)
> Does reordering the keys on a keyboard allows entering text more
No, that is why I think one can just as well use ABCDEF. QWERTY and
friends reordered it, according to rumor, for various reasons, like
slowing it down on mechanical typewriters, or whatever.
> Or even allows to enter new missing characters?
I am not sure what you mean here. There are some 100000 plus Unicode
characters to be entered.
> I don't think so.
So we agree.
> People are used to some general layout, and there are even
> standard practives that apply to ALL keyboards for the position of
> within groups.
So then just keep on using the keyboard maps you are used to, if you
so like. It odes not have anything with the design of new keyboard
maps to do.
> But then, why would you need a "Unicode" keyboard?
I already told you: For example, there are thousand of mathematical
characters those that type mathematics want to use. Designing ones
own keyboard maps for such a large number of charcters is very time
> I think that such thing
> is just a dream, that can never occur because the first ficus of a
> is not to allow input any character, but real texts in some language
> (s) the
> most efficiently as possible for the target languages.
This does not apply to the current keyboard layouts, like the QWERTY,
that was developed for rather ad hoc reasons.
> Keyboard designs need first to be efficient for the task we assign
> to them:
> input text, rather then inputting characters.
There needs not be a one single keyboard map.
> But reordering the alphabetic groups on keyboards will not make things
This is what I am saying: ABCDEF is much better than QWERTY.
> You'll immediately notice that usability if the first thing
> considered, and that different layouts are needed depending on both
> target languages and the user's limited capabilities (think about
> keyboards for single-handed persons, or about other layouts needed for
> people that can type but must input text using a stick in their
> mouth): the
> frequency of characters that need to be composed must be considered.
> So I don't think that any unique "standard" layout will match all
> needs. And
> a pure "alphabetic" layout is not the best in all situation. We
> need several
> layouts according to usage.
A "universal keyboard layout" may in fact be build by a set of
keyboard maps. For example, one could have one used for all Latin
latin, one for all Cyrillic scripts, and so on. Then add more
specialized keyboard layouts for specific languages, if those already
in use are not sufficient. If one for one reson does not know the
specialized language keyboard layout, one can use the more general one.
> In fact, even the so-called "alphabetic" keyboards are not purely
> the binary ordering of codepoints,
Look at the numbers of the code points of the Latin mathematical
semantic styles: some of them are quite spread out. If one wants to
enter the code points hexadecimally, Mac OS X already has such a
This comes to the core of the problem: A keyboard layout must be
logical to the human. THis is difficult to achieve. This is what
takes time, if one should make ones own keyboard maps.
> which is unidimensional, when all
> keyboards have several dimensions (at least 2, but supplementary
> are created by assigning multiple characters on the same key and
> * using multiple input modes controlled by other keys (modifiers,
> * using various contextual input methods (including repeating the
> to select the next character, on limited keyboards like mobile phones)
> * using programmable keys (controlled by user preferences or by
> There will never be any "Unicode" keyboard to input text by selecting
> characters within a giant binary ordered matrix; even a Unicode
> map on display needs to allow selection of subsets to display only
> a part of
> the matrix. This tool is only convenient for selecting characters
> occasionally, but nobody wants to use it to enter text efficiently.
It already exists: Mac OS X has a "Character Palette" which in
essence is doing just that. As you note, it is not convenient for
typing, but only some uses.
So one needs a set of input methods for different contexts.
This archive was generated by hypermail 2.1.5 : Sat May 26 2007 - 05:18:03 CDT