[lsb-discuss] regarding ncurses test cases

Theodore Tso tytso at mit.edu
Fri Jun 30 05:56:02 PDT 2006


It's good to hear that ISPRAS is already working on this!

On Thu, Jun 29, 2006 at 07:51:44PM +0400, Alexey V. Khoroshilov wrote:
> Drawback of the approach is a restricted set of terminal types, which could
> be tested.
> Because for each terminal type we have to have an emulator supporting dump
> of the current state of the virtual screen to a file.

This isn't necessarily a major drawback, I would think; the question
is what is being tested, the ncurses library or the terminfo/termcap
database?  As long termcap database is valid for the vt100, such that
escape sequence defined by dl1/dl really does delete a line on the
vt100, and the string defined by cub1/le really does move the cursor
left, and the vt100 terminfo entry is a reasonably complete, it should
do a pretty good job testing the ncurses library.

This of course is no guarantee that the Wyse terminal definition is
correct, or that perhaps some obscure terminal uses some
terminfo/termcap capability that wouldn't be tested using just a vt100
emulator, but it seems that this is what a code coverage tool would be
good at validating, and in any case, 99.99% of the what gets used
today is a vt100 derivitive, including the current X windows
terminal/konsole programs.


						- Ted




More information about the lsb-discuss mailing list