Reading initialization/logout files again.

Reading initialization/logout files again.

[Peter Stephenson]

This was some discussion about this before.  The latest proposal to allow
more flexibility as follows:

- The NO_RCS option (or negated RCS option) remains as before, i.e. it only
takes affect for initialisation files if set as a starting option to the
shell or in /etc/zshenv, but if set on logout it disables sourcing
/etc/zlogout and .zlogout too.

- A new NO_GLOBAL_RCS option.  For greater flexibility, you could set or
unset this anywhere and it will determine whether the /etc/z* files are
sourced at any stage.  As before, /etc/zshenv is always sourced since there
are legitimate reasons for this (as well as illegitimate ones).

- The previously proposed option GLOBAL_RCS_FIRST would be removed.

There are alternatives, however:

1. it's not well documented that setting NO_RCS prevents running
/etc/zlogout and .zlogout.  So if no-one was setting it in their .z* files
for this purpose, it would be possible to make the NO_RCS option more
flexible and allow it to determine the subsequent behaviour at any stage,
as NO_GLOBAL_RCS would do.  But if people are (say) setting it in .zprofile
in the knowledge that it doesn't take effect until logout this will cause
problems, and anyone who is should say so now.

Currently, altering it for other reasons in
.zshenv/.zprofile/.zshrc/.zlogin doesn't have any effect, so we probably
don't need to worry about that happening.  This is the cleanest option.
I'm not sure why NO_RCS behaves the way it does at the moment; it may be so
that the administrator can unset it in /etc/zshenv and so enforce the rest
of the scripts, but it should be quite enough having one file that is
always run.


2. for consistency, if people didn't really care about greater control, it
would be possible to make NO_GLOBAL_RCS work just like NO_RCS, except that
it applies to files in /etc only and is tested just after .zshenv is run.
It's not that consistent, though; as the idea is to let the user get their
hands on it, it has to be tested later than NO_RCS anyway.

-- 
Peter Stephenson <pws@ibmth.df.unipi.it>       Tel: +39 050 844536
WWW:  http://www.ifh.de/~pws/
Dipartimento di Fisica, Via Buonarroti 2, 56127 Pisa, Italy

Re: Reading initialization/logout files again.

[Bart Schaefer]

(I feel a bit odd being the poster of ten consecutive zsh-workers article
numbers and several more on zsh-users, but I'm still catching up after my
system downtime.)

On May 23,  4:41pm, Peter Stephenson wrote:
} Subject: Reading initialization/logout files again.
}
} This was some discussion about this before.

In fact, very far before.  Start at
      http://www.zsh.org/cgi-bin/mla/redirect?WORKERNUMBER=488
(yes, four hundred eighty-eight) and work your way up the references links.

} I'm not sure why NO_RCS behaves the way it does at the moment; it may be so
} that the administrator can unset it in /etc/zshenv

Also note that in the thread around zsh-workers/1414, Zoltan ignored me when
I repeatedly asked why NO_RCS works the way it does.

} The latest proposal to allow more flexibility as follows:
} 
} - The NO_RCS option (or negated RCS option) remains as before
} 
} - A new NO_GLOBAL_RCS [...] will determine whether the /etc/z* files are
} sourced at any stage.
} 
} - The previously proposed option GLOBAL_RCS_FIRST would be removed.
} 
} There are alternatives, however:
} 
} 1. [...] make the NO_RCS option more
} flexible and allow it to determine the subsequent behaviour at any stage
} 
} 2. [...] make NO_GLOBAL_RCS work just like NO_RCS, except that
} it applies to files in /etc only and is tested just after .zshenv is run.

All three of these suggestions (latest, #1, #2) are OK.  I slightly prefer
alternative #1, and slightly dislike alternative #2.

-- 
Bart Schaefer                                 Brass Lantern Enterprises
http://www.well.com/user/barts              http://www.brasslantern.com