Darren,
I think I'm lost. Due to starting these four fasttracks with
Subject lines that didn't contain the case numbers, it looks like a log
of the discussions on them never made it into the case logs. Is there
any way for you (perhaps with John's help) to recover the lost
commentary?
>Date: Wed, 25 Apr 2007 12:36:34 -0400
>From: James Carlson <james.d.carlson at sun.com>
... ... ...
>
>Darren J Moffat writes:
>> - The settings of MANPATH and PAGER are set in "/etc/profile" for
>> the whole machine but users are always allowed to override these settings
>> in their local environment files (e.g. ~/.profile, ~/.kshrc, ~/.bashrc)
>> on demand.
>
>The current /etc/profile sets only TERM and exports LOGNAME and PATH.
>
>Are there any standards conformance issues with adding new default
>values and thus new behaviors? (Don ... ?)
I don't see that the standards specify who is allowed to initialize
PAGER, but once it is initialized if affects at least mailx(1) and
man(1). SVID3 says that if PAGER is not set, mailx and man use pg(1).
POSIX/SUS say that if PAGER is not set, mailx and man use more(1) or
some other utility documented in the system documentation. If we set
PAGER to less as part of this project, the utility and the less man
page must also be available on all affected systems. It appears that
less doesn't come from the ON consolidation and the less(1) man page
doesn't have an ATTRIBUTES section saying what package it comes in; is
it in a package that is always installed?
MANPATH is only documented in the standards to point out that it exists
on some implementations and should not be used for any other purpose.
It was intentionally left out of the standard with the following
rationale:
"The historical MANPATH variable is not included in POSIX
because no attempt is made to specify naming conventions for
reference page files, nor even to mandate that they are files
at all. On some implementations they could be a true database,
a hypertext file, or even fixed strings within the man
executable. The standard developers considered the portability
of reference pages to be outside their scope of work. However,
users should be aware that MANPATH is implemented on a number
of historical systems and that it can be used to tailor the
search pattern for reference pages from the various categories
(utilities, functions, file formats, and so on) when the system
administrator reveals the location and conventions for
reference pages on the system."
>
>Should these changes be reflected in environ(5) as well?
Absolutely, if these changes are made.
