Peter said:
>2. How do I get software X to know how to process my PUA characters, or how
>do I document my characters for others to understand my data?
Michael replied...
> In principle it would work, if the OSes are being written to handle user
> editing of such things. Ten euros sez they ain't.
Well, at one time when I believe it was possible with the precursor to Mac
OS X to change the behavior & properties in the "Cocoa" text system. It's
merely a matter of loading or re-loading the data tables. I don't know if
the data format is published...
In an object-oriented system, you could pretty easily over-ride only part
of the data dealing with the PUA by funnelling property requests in the PUA
range to a set of user-functions, or to user-loadable data blocks. (On
Mac OS X for instance, the "User Defaults" mechanism should be easily
adaptable with load/unload hooks for this and would work on a per-machine,
per-user or per-app basis, or all three.)
Mark Leisher's publicly available "UCData" project also allows re-loading
the property data with explicit load and unload functions. Overloading the
utility functions to filter PUA codes to another data set would be
in-principle rather easy. E.g., I mean something as trivial as this at the
top of _ucprop_lookup():
if (_userDataLoaded && is_PUA_Char(code)) {
return _ucprop_lookup_PUA(code, n);
}
and providing a set of load functions for user data. This project code
already uses the Unicode Data file as-is for its initial input. PUA data
could be used on a per-installation basis by simply adding data to the data
file.
Rick
