Re: cvs-1.11.13 test failure (basic2-64)

[Paul Jarc]

lawrence.jones@ugsplm.com (Larry Jones) wrote:
> Paul Jarc writes:
>> I'd say a path containing symlinks is just as "actual" as any other,
>
> You're entitled to your opinion.  But you're wrong.  :-)

IME, things usually break on symlinks only when they go to extra
effort to handle symlinks specially, and get it wrong.  Most code can,
and should, accept any filename given to it without knowing whether
symlinks are involved.

> In general, the tests depend on the previous tests in the same group
> having completed successfully, so the test suite always stops on the
> first error.

What about just skipping to the next group, then?  The failure should
still be remembered so the exit status would still be nonzero, but any
unrelated later tests could still be performed.


paul