"YH" == Yannis Haralambous writes:
YH> I think xindy could do much better: consider the case where you
YH> have Spanish and Greek, or French and Russian, or Italian and
YH> Hebrew, or German and Polish... there are so many of
YH> combinations of two languages having different 8-bit codepages.
YH> There is no way to say to xindy that we are switching codepages
YH> and sorting must be different for each one. Fatally xindy will
YH> identify the ^^e0 of one codepage with ^^e0 of the other codepage
YH> and the result will be crazy.
YH> The most natural solution to this is to work in 16-bits. Everyone
YH> is switching to 16-bits today, even Microsoft (which I hate so
YH> much) with Windows2000.
Another natural solution is to use inputenc package (when using 8-bit
codepages in "original" TeX; i like Omega's universal OTP concept). :-)
For example, in case of cyrillic languages, when one uses 8-bit
codepages wiht inputenc, the AUX files will not contain such things as
^^e0 (for which it is impossible to guess the charset; thus AUX files
must not contain raw 8-bit chars), but will contain commands like
"\IeC {\CYRA }" for each cyrillic letter; and xindy will happily use
this information to provide correctr ordering (the same is valid for
all other 8-bit non-cyrillic codepages provided that inputenc is
used).
Best regards, -- Vladimir.