Re: sanity failure

[Paul Edwards]

"Larry Jones" <lawrence.jones@eds.com> wrote in message 
news:mailman.7710.1055344395.21513.bug-cvs@gnu.org...
> Paul Edwards writes [quoting Derek]:
> > >
> > > Did it tell you why it thought expr was broken?
> >
> > Not sure what you mean.  It said that I should make a GNU
> > expr available if I didn't want the spurious passing (not
> > failing) of tests.
>
> The warning message usually includes some indication of what the problem
> is, such as not handling multiple lines or long lines.

Warning: you are using a version of expr which does not correctly
match large patterns.  Some tests may spuriously pass.
You may wish to make sure GNU expr is in your path.

> > Anyway, I tried sticking /usr/xpg4/bin at the front of my PATH
> > today, and got rid of the ID and EXPR variables, but it didn't
> > make any difference.  I still got a spurious fail when I was
> > promised spurious passing.  :-)
>
> Spurious passing is by far the more serious problem since you'll think
> everything is fine if all the tests pass and that may not be true.

This is true from the build point of view.

> That's why we explicitly warn about it.  Failures have to be
> investigated whether they're spurious or not.  :-)

But from a user point of view, 10% of the tests got run before
I got some crap message.  So I don't run sanity, because it is
producing incorrect results, CVS is not producing incorrect
results.

Spurious passing of that test would have allowed me to get
past that test and finish 100% of the suite, albeit with a few
test cases that were essentially skipped.

It's a choice between running sanity once, on 10% of cases,
and then never running it again because I "know" it doesn't
work, or running sanity on 90% of cases every time because
it can at least detect genuine errors in that.

But so long as CVS compiles on my system, all's well that
ends well.  :-)

BFN.  Paul.