Re: Changes in latin-ltx.el

[Dave Love]

>>>>> Eli Zaretskii writes:

 > If you want latin-ltx.el to include all of the Hebrew characters, I
 > can only welcome that.  

I wish things weren't forbidden one day and welcome another.

 > But then please (a) provide _all_ of the Hebrew characters, not
 > just the first 4, and (b) please provide a way to type the Hebrew
 > letter-like math symbols (Aleph to Daleth), 

As I said, people are welcome to add them if they want them and
they're not confusing to a user.  I don't know TeX names for any other
letters (and I have the LaTeX symbol list).

 > because those are different characters in Unicode.

I don't immediately care about other compatibility characters either.

 > Those who don't know will randomly choose one of the two
 > alternatives, with no more harm 

It's hardly greatly harmful, but they won't find the character for
which they searched.

 > than if you offer them only one (randomly chosen by you)
 > alternative.

It isn't randomly chosen.

 > Those who _do_ know will choose the right alternative; those are
 > the ones I care for and want to cater to.

I was catering for a fairly expert user.  Quail doesn't provide any
means that I know to distinguish input characters with similar or
undisplayable glyphs.  It's not a very good solution for what I wanted
to do, but it's the only one on offer and I've talked about better
ways.  Why not _try_ this sort of thing?  It's hardly confined to
compatibility characters.

 > The mnemonic we define for each one of the alternatives might help
 > them make the right choice;

I said \...sym is perfectly fine, apart from the fact that Quail
doesn't offer completion, so that you can see the complete
alternatives.

 > they can also use "C-u C-x =" for more detailed info.

Even with extra info I've added, it's unclear to me how that would
necessarily help, and such feedback isn't available when entering
Quail sequences.

 > Yes, having two characters that are almost identical is a mess.  But
 > for better or for worse, Unicode does define two different codepoints
 > with slightly different attributes in this case, and we have no good
 > alternative but support both of them.  

I've contributed two methods for general input of unicodes and I don't
feel compelled to add every Unicode character to latin-ltx even if
there was a recognized TeX command for them.

 > The bidi reordering produces a terribly messed up display when you
 > mix Hebrew letters with symbols and math operators; 

Then it sounds as though there's something wrong with it.  Yudit
displays what I expect it to, not messed up.

 > anyone who have seen that mess will I think understand why we need
 > to distinguish clearly between the right-to-left letters and the
 > left-to-right symbols.

I'm quite clear about the distinction.