Martin Schwartz writes:
: you are most certainly aware that we've currently two modules doing
: character mappings to and from Unicode. This is Unicode::Map8 for 8 bit
: character sets done by Gisle Aas and Unicode::Map by me basically able
: to do any n->m bit conversions. The obvious disadvantages of this
: situation are:
:
: * One needs to install two modules doing similar tasks
: * The modules have a (slightly) different syntax
: * We've two different map file structures
: * Map files are not synchronized with related projects, especially
: Tcl.
:
: Especially Unicode::Map has these problems:
:
: * Character mappings are based on the mappings of Unicode Version 1.1
: AND some at this point better quality / different map files from
: another destination
: * The "binary" mapfile is complicated at places where it's not needed
: * The module applies the module "Startup". I had quite some plans
: with this a while ago but basically it is not needed and confuses.
:
: Gisle's Unicode::Map8 implementation very straight. Basically
: Unicode::Map8 builds perl tables to map, whereas Unicode::Map uses Hashs
: to map. Unicode::Map can therefore easily handle 16bit->16bit or
: n*8bit->m*8bit conversions, Unicode::Map8 would need some clever
: extension.
Whatever tables we come up with, it would be nice if they could be in
a form that is usable by built-in utf8 encoder/decoders we're planning
to attach to Perl streams at open time (or later). So in a sense, we
need to turn two modules into zero modules...
Larry