[pdftex] Re: [tex-live] problems with non-final sigma

[Robin Fairbairns]

Tigran Aivazian writes:

> No, it is something more fundamental, like catcode redefining or something 
> like that, which I don't fully understand (but I am sure someone does).

doesn't sound like it to me (and indeed it doesn't occur in the
absence of babel greek of any sort -- nb i only know modern greek).
it would take rather extreme coding to decide to change the catcode of
a letter.  far more likely is a boundary character ligature, as
vladimir suggests -- except at a cursory glance (i've installed the
fonts since i started typing ... helps to maintain a ctan mirror at
home, and a sunday morning to play) i can't find the ligature in the
table.

however, while i see your problem, i don't really understand your
premise.  the only circumstance i can imagine needing a non-terminal
sigma is in an instructional manual, and in that case it's easy enough
to do a fiddle like yours with alpha (though i personally wouldn't
code it like you did...).

> So, so far, there seems to be NO way to force a non-final sigma in a Greek
> text, therefore I suggest to remove the context-sensitive logic from
> Babel/Greek language definition because it introduces this serious bug.  
> I.e. unless someone does come up with a working way to display a non-final
> sigma.

my guess is there's no context sensitivity in babel greek (for sure,
none is needed in this area) -- it's all in the font.  and there is
notoriously no way to switch off ligature effects in tex.  :-(

incidentally, knuth's test \lowers, in testfont, is particularly
useless in this case, since it would appear not to allow any two
letters to ligature -- every sigma in it appears as a terminal sigma.
even \names only produces one medial sigma that i can see (which has
me boggling about how he implements that test ... i can imagine such
an infelicity in \lowers).

robin

nb, i've placed reply-to texhax in the header.  *please* don't post
general tex questions to pdftex or tex-live -- this isn't a pdftex or
a packaging effect, which are what those lists are for.