How to type sporadic Unicode (was: User interface for keyboard input)

From: Martin Kochanski (
Date: Fri Jul 19 2002 - 05:48:29 EDT

Thank you! I was starting to write a detailed reply to Otto Stolz but it was beginning to sound unnecessarily negative; so I'll fold my response to his email into yours.

At 22:02 18/07/02 -0700, Doug Ewell wrote:
>I suspect it will be a much greater benefit to the widespread,
>intentional use of Unicode by ordinary users (e.g. to enter the
>occasional em-dash or Greek alpha) than many people imagine.

I wish I had put it that well. I've re-titled this thread to narrow the focus.

If you have a need to type Greek and install a Greek keyboard layout, then I'm not going to give you any special support - because I don't have to: it all just works. In the more general case, if you set up *any* kind of special keyboard layout, and that matches the characters that you want to type, then of course it will just work, and again I don't need to give you any help.

Suppose, though, that you have been asked to enter some data into (say) a bibliographical database. Greek isn't too bad - all you have to do is recognise that it *is* Greek, ask your systems people for the Windows CDROM so that you can install a Greek keyboard, install it, switch to it, and hit every key on the keyboard until you find the one that generates the right symbol (alpha was too easy: try phi). As soon as you start entering Greek on a regular basis, you'll get used to it-- Unicode entry is no longer sporadic and therefore no longer a problem. But now: if you have to type a w with a circumflex on it, or an L with a slash, what keyboard layout do you load and what key do you press?

We are somewhat restricted because as a matter of policy, we write software for people to run under Windows - that is, not for Windows 95, Windows 98, Windows ME, Windows NT, Windows 2000, or Windows XP. Consequently, CHARMAP isn't a solution, because on some Windows systems it doesn't always display more than 224 characters. I do appreciate Otto Stolz's use of "contemporary Windows", but for us "contemporary Windows" has to mean *not* "versions of Windows currently being sold" but "versions of Windows currently being used", and since some of our customers are academic instutions and charities, "contemporary Windows" included all the operating systems I listed, with the possible exception of Windows 95.

Alt+X would have been a solution if it had been consistently implemented: but there are several different and incompatible implementations floating around. The fundamental problem is, as you pointed out, that CACF9 AltX could have five different meanings and there is no way for the software to guess. Even within Microsoft's own software, there are wide variations in behaviour (convert all previously-typed characters; convert all characters before the current cursor position; convert highlighted characters only); and there is sometimes an inverse operation provided (Unicode character to hex), but that is sometimes Alt+X and sometimes Alt+Shift+X.

ISO 14755 looks promising. For those who don't make it their daily reading (or to show off my pitiful misunderstanding of it), it could be described as "Use the Alt+nnn approach, but use Ctrl+Shift instead of Alt, to indicate hexadecimal digits". Am I right about this? Does any real software implement it? And how is a Windows program to cope if it already uses something like Ctrl+Shift+C as a keyboard shortcut?

This archive was generated by hypermail 2.1.2 : Fri Jul 19 2002 - 04:02:43 EDT