I think we are talking about two different issues here. It's important to keep these separate, to avoid talking past each other.
Mats Blakstad wrote: > After managing to add the keyboard to XKB I started on a new venture > of trying to make a windows version of the keyboard using this: > > [link to Microsoft Keyboard Layout Creator] > > It is nearly impossible to replicate as it seems like you can only add > dead keys if they have a precomposed character. Mats is talking about the fact that a dead key combination (of any length) under Windows can generate only a single UTF-16 code unit. This is a Windows architectural limitation, and cannot be fixed by updating MSKLC. It can only be circumvented by using Keyman or another third-party solution that runs at a layer above the Windows architecture. Philippe Verdy wrote: > My opinion is that MSKLC should be updated to support chained dead > keys (internally they are supported by the OS), using more keyboard > maps. The fact that MSKLC does not support chained dead keys is perhaps related to the problem Mats is experiencing, but it is a different issue. Even if MSKLC were updated to allow chaining of dead keys, Mats still could not use this capability to type a TILDE dead key, then an ACUTE dead key, and then an EPSILON key and get "Ɛ̃́" LATIN CAPITAL LETTER EPSILON WITH TILDE AND ACUTE. The reason, as Mats said, is that the NFC form of this double-accented letter is still 3 code units in length, 2 more than the Windows architecture supports. Furthermore, even though many of us would like for MSKLC to be updated, the reality is that its developer (Michael Kaplan) is no longer with us, and Microsoft had already terminated MSKLC development (a source of frequent frustration to Michael). We can all wish that Microsoft would reverse itself and start devoting resources to this project of Michael's, but it's probably not going to happen. A more realistic course of action might be for someone outside of Microsoft, maybe someone on this list, to create their own GUI wrapper around the Microsoft engine, a "new MSKLC" so to speak. That new project could remove the MSKLC limitation, but not the Windows one. Mats wrote: > So I wonder if it could be a solution for a precomposed double tone? > So one unicode for tilde+acute and another for tilde+grave? If Unicode policy is what it used to be, then Philippe is correct: vendor limits are not an adequate justification for encoding double diacritics. Doing so would introduce new ambiguities, just like encoding new precomposed versions of characters that already have decomposed representations. Denis Jacquerye suggested using the letter as the dead key instead of the diacritic. Perhaps a more straightforward approach would be to give the diacritical marks their own normal keys, so the user could type EPSILON, (combining) TILDE, (combining) ACUTE. Marcel Schneider's suggestion of using Keyman instead might be the best, if it is mandatory for the Windows version of this layout to be identical to the Ubuntu version, for reasons I don't understand (many keyboard layouts are already not constant across platforms). -- Doug Ewell | Thornton, CO, US | ewellic.org
