At 11:13 PM -0500 3/3/01, Charles Lane wrote:

>One early difficulty was that I normally build and test Perl with a
>rooted-logical environment. I was getting test errors (not from the
>directory depth), but tracking down the other errors by running the test
>manually from a non-rooted environment exposed the problems with
>directory depth.
>
>Of course, running manually meant I also had "PERL_ROOT" defined, so
>Perl had no difficulty finding its libraries.  Damn.

Been there, done that.  I think it could also grab modules from the
library path of your installed Perl in this situation.

>So, a couple of  possible solutions: move the test directory stuff to
>somewhere higher in the directory heirarchy...I think we're okay with
>one level below the directories where the test scripts live.
>
>Do the "rooted logical" so that it corresponds to the usual PERL_ROOT.
>That way the relative directories will work properly.

Moving higher in the directory tree and/or keeping find.t's private
play space shallower seems preferable to changing the current
directory.  Something like your rooted logical code could probably be
made to work but it couldn't be in the BEGIN block; it'd have to be
after any libraries are loaded, and it would break as soon as someone
added something elsewhere in the test that depended on being in 't'
(a fairly reasonable assumption).

Michael, are there other reasons your patch was withdrawn or are we
the only ones causing you trouble?
-- 
____________________________________________
Craig A. Berry
mailto:[EMAIL PROTECTED]

Reply via email to