>
> Seeds, Glen <yyyyyyyyyy@xxxxxxxxxx> wrote, on 29 May 2003:
> >
> > From a practical standpoint, it's extremely useful to be able to distinguish
> > between "does not exist (or exists but is not a file)" and "an error
> > prevents determining whether it exists or not". Since most implementations
> > can't return -1, they have to be able to return 1 and something >1 to
> > distinguish these. It sure would be nice if the standard did not preclude
> > this.
>
> No, it would not be nice. If any system actually implemented "test" this
> way, a heck of a lot of existing shell scripts would not work properly on
> that system. We *must* clarify the standard so that it requires the
> historical behaviour.
>
> --
> Geoff Clare
> The Open Group, Apex Plaza, Forbury Road, Reading, RG1 1AX, England
> Email: yyyyyyy@xxxxxxxxxxxxx Tel: +44 118 9508311
>
Agreed. The reason why this came up in the first place is that the
standard does not clearly dictate the historical behavior when pathname
resolution errors are encountered during processing of the test utility.
I've claimed in this space that the standard actually dictates something
different from the historical behavior, but I'll be satisfied with a
clarification that test does not consider pathname resolution errors as
errors in the function of the test utility.
--
Marc Stephenson IBM Server Group - Austin, TX
Internet: yyyy@xxxxxxxxxxxxxx
Phone: 512-327-5670 T/L 678-3189
|