termcap strangeness

[Guy Harris]

>	I suspect that, knowing these "features" of common terminals,
>that the writers of the termcap/curses package punted and returned 79/23
>whenever they saw 80/24 -- that way they could assure that the screen
>would look reasonable under vi/emacs/whatever.  Any veteran wizards know
>different?

Yup, I know that neither the BSD "termcap" nor the SunOS "termcap" does
any such thing.

The SunOS "termcap" *does* do a TIOCGSIZE on older releases, and either
a TIOCGSIZE or TIOCGWINSZ on later releases, to see whether the tty
driver has a notion of the screen size that differs from the notion in
the "termcap" entry (for example, the "sun" entry is 34 rows, 80
columns, but a "shelltool", which uses that "termcap" entry, is not
necessarily 34x80) and, if it is (i.e., if the sizes are non-zero), it
returns the sizes it got from the TIOC call instead. 

The Apollo "termcap" code may do something similar.