On 25 Jun 2008, at 8:47 pm, Mojca Miklavec wrote:
Hello Jonathan (or anyone else on the list),
Can you please tell me if the way we load patterns for French, Latin
and Danish is likely to work the same way as it did with old patterns?
I have no idea how to test that.
(Talking about the code in svn repository.)
I have not actually tested it but it looks to me like it should work.
I also prepared some extracts of apostrophes (represented twice in
UTF-8), but did anyone take a look at Petr's suggestion?
To use \savinghyphcodes, that is? I have never tested this in xetex,
and I suspect nobody else has used it either! AFAIK, it should work,
but only for the \lccodes of chars < 256; it has not been extended to
cover all the Unicode character codes.
This means that you could try the following approach:
* use U+2019 (the "correct" Unicode apostrophe) in any patterns that
require it, rather than U+0027
* when loading the patterns in a Unicode-capable engine, set
\lccode"27="2019
\savinghyphcodes=1
* when loading the patterns in an 8-bit engine, use active chars to
map the apostrophe to 0x27, and set
\lccode"27="27
The "reverse" strategy of setting \lccode"2019="27 for the patterns
can *not* be preserved by \savinghyphcodes in xetex.
I have *NOT* tested this; I'll try to do so soon, but if someone else
wants to experiment first, that would be very welcome.
JK
