Re: Comments on FHS testsuite run
On Sat, Oct 21, 2000 at 02:54:12PM +1100, Christopher Yeoh <email@example.com> spake forth:
> > I agree that tcl shouldn't be required on an install. On the other hand, I
> > don't mind making a symlink so that when it *is* installed, /usr/bin/tcl
> > works - assuming this isn't a bug in the testsuite or the standard. I've
> > never seen it as /usr/bin/tcl on any system ever...
> I believe that its a bug in the specification and should be
> /usr/bin/tclsh (the test suite just tests what is stated in the FHS
> spec). Having a symlink that doesn't link to anything won't work as
> the test attempts to execute the program.
Right. I mean, I'd be willing to put a symlink from /usr/bin/tclsh (managed
by alternatives) to /usr/bin/tcl *if* that's what it should actually be, but
I suspect that it's an error in the spec as well.
> There a few areas in the specification where instead of <foo> being
> required, it should be conditional on <bar> being supported. This is
> already done in the FHS spec for some things - eg. /var/account only
> being needed if process accounting is supported.
Agreed. I guess that's why it's not yet finalized.
Mike Markley <firstname.lastname@example.org>
PGP: 0xA9592D4D 62 A7 11 E2 23 AD 4F 57 27 05 1A 76 56 92 D5 F6
GPG: 0x3B047084 7FC7 0DC0 EF31 DF83 7313 FE2B 77A8 F36A 3B04 7084
I've already got a female to worry about. Her name is the Enterprise.
- Kirk, "The Corbomite Maneuver", stardate 1514.0