Re: Changes in latin-ltx.el

[Richard Stallman]

    I think it actually could be confusing the way it is now: someone who
    uses a TeX-like input method will most probably type scientific text,
    not Hebrew text.

That seems likely.

                      That someone will not expect Hebrew characters in
    their text instead of mathematical symbols.

A priori I suppose he would not care which one he gets, if they look
and behave the same.

    No, please don't!  The TeX-like strings that latin-ltx uses are
    supposed to produce mathematical symbols, isn't that true?  If so,
    it's simply _wrong_ to produce Hebrew letters instead, because in
    Unicode, these are different characters.  For example, they have
    different bidirectional properties, so when bidirectional display will
    be supported, the mathematical expressions, in which these characters
    will be used, will be messed up by the bidi reordering algorithm!

This seems like a significant difference in behavior, and a strong
argument for reverting the change.  Is there a counterargument?