On Mon, Nov 13, 2006 at 08:57:37PM +0100, Eric Y. Kow wrote:
> > Then we could make this part of the standard test suite by renaming it
> > with a .sh at the end.
> 
> Hmm: not sure that this would work well in the shell harness.  Part of
> the point is to test interactions between ssh and the user, whereas the
> harness supresses the output as far as I know.

Hmmmm.  I must admit that I've only glanced at this.  But it does seem like
with a proper ssh configuration we should be able to run some "real" tests
of the ssh functionality without user input.  And the output is something
that we can test easily enough in an automated manner.  It may not test
what you were looking at, but should be able to test *something*, and
that's more than we're currently doing.  True, ssh is a commonly used
feature that's unlikely to be accidentally broken, but I'm a big fan of
getting our testing as close to comprehensive as possible.  If nothing
else, it makes it easier to find simple cases that trigger (new) bugs.  And
once the framework is in there, adding new test cases (for example, that
illustrate problems that will invariably show up when switching to
System.Process) should be relatively easy.

It may be that we'd have to disable tests that require user input when
running under harness...
-- 
David Roundy
Dept. of Physics
Oregon State University

Attachment: signature.asc
Description: Digital signature

_______________________________________________
darcs-devel mailing list
[email protected]
http://www.abridgegame.org/cgi-bin/mailman/listinfo/darcs-devel

Reply via email to