Charles Lane wrote:
> If you look for VMSPIPE and can't find it at all, what do you do?
> You can't run system(), you can't do piped i/o. I'd say that a -F-
> is an appropriate response to that kind of failure, maybe a *bit*
> on the harsh side.
There are plenty of things I can do with perl with out system or piped
I/O. perl -e "print ""Hello World\n"";" plus oodles of regular filehandle
I/O examples come readily to mind. In fact I'd go so far as to assert
that most of what I use perl for does not entail piped I/O or system() stuff.
> > Note that one of the reasons that rpm was rewritten in C was so that a
> > floppy based install did not have to have a perl binary lying around.
> > It'd be nice if perl could regain some of it's little portable utility
> > character (er, maybe :-).
>
> Perl? Little? Surely you jest!
I do.
> So let's see: PERL_ROOT might not be there, the Perl executable itself
> may be on a floppy or elsewhere. So that leaves the "write a tempfile"
> solution.
>
> Maybe as a last-gasp, everything-else-fails, solution.
>
> BTW, the reason I suggested that putting VMSPIPE.COM in [.t] is
> possibly a bad idea was not that it wouldn't work, but that there are
> tests (that I've run afoul of before) that check for files in [.t] and
> bomb if they don't see what they want, or if they find more than they
> want.
The glob tests? Just curious. I think these can be modified if they
rear their ugly heads in the future.
> I don't expect this to be a "real" problem, but it does highlight the
> difficulties of operating in a "no PERL_ROOT" environment.
I have no doubt that getting it to work without PERL_ROOT is difficult.
But `mms test` already does this.
Peter Prymmer
