CGJ is NOT made to create (or even hint) ligatures ; and certainly not in this context.
Anyway, there's not need for such ligatures in Breton, except in very limited contexts (even in crosswords, you would encode the digram CH or trigram C’H directly in each square, even for vertical words, but even in this context, I've never seen any crossword in Breton where digrams or trigrams needed to fit within the same square, and the apostrophe is implicitly dropped in such limited rendering, just like other distinctions of accents or letter case; not different from crosswords in French). Digrams and trigrams can correctly be handled by tailored collation rules, they do not require specific encoding, but most certainly only within decorated font styles, where many other ligatures will be present as well. Yes, there may exist contexts where you'll want a graphic ligature for the digram CH or trigram C’H, but why would it be visually distinct from separated letters ? Even in monospaced fonts, you'll use the regular letters C and H and an apostrophe (the only problem being the effective encoding of the apostrophe, as the ASCII vertical quote is most frequently found, even if the curly right apostrophe is probably better typographically : this is not different in French or even English or any other Latin-written languages ; it is not really ambiguous, even for collation purpose where those apostrophes will collate together as equivalent up to the third level, i.e. level 1 for base letters and digrams/trigrams considered single letters, level 2 for accent differences, level 3 for case differences, with the last implicit level for binary differences of encoded code points, including canonical equivalents). For this reason, you should not even have to generate any joiner control on a normal Breton keyboard layout, even if you provide convenient keys for typing those digram/trigram. The cultural aspect of the traditional alphabet is only a mattter of collation, not rendering. For this reason, most Bretons just type on standard French keyboards without modification. You may still map an extra key for the preferred left curly apostrophe (you can do that as well for extending the French keyboard) ; but mapping ZWJ on a key will be only useful for typesetters that want explicit ligatures on final documents ; it is not needed for almost all usual texts. Such mapping of ZWJ could in fact be used anywhere else where they would be graphically interesting (for example in "ct", "st", "ff", ""ffi", "ffl", even if some fonts will provide these ligatures automatically). If you don't want automatically generated ligatures, you will want ZWNJ. Both ZWJ and ZWNJ will be ignored in collation. -- Philippe. 2011/6/29 Richard Wordingham <richard.wording...@ntlworld.com>: > On Wed, 29 Jun 2011 03:49:42 +0000 > Peter Constable <peter...@microsoft.com> wrote: > >> From: unicode-bou...@unicode.org [mailto:unicode-bou...@unicode.org] >> On Behalf Of Jean-François Colson >> >> > * In the C’HWERTY layout on Linux, the digraph and trigraph had to >> > be replaced by six PUA characters >> >> That would appear to be a limitation of the input method. > > It is indeed a limitation of X. I get round it on Ubuntu by using > IBus and KMFL (Keyman for Linux), which then allows me to use dead keys > for sequences, something which is (or used to be) beyond MSKLC. > >> > * Since those two letters must be encoded in 2 or 3 characters, >> > with a monospace font, they are twice or 3 times larger than the >> > other letters. >> > >> > To solve this last problem, would it be possible to make a font in >> > which c ZWJ h would be displayed as a new glyph? >> >> That is certainly possible. > > However, what you probably need is a 'monospace' font where c h is > displayed as a ligature and c CGJ h is not! Incidentally, you can even > use KMFL to generate c CGJ h for the 2-key sequence c h. > > Richard. > > >