On Sun, 8 Dec 2002 03:53:00 +0200 (IST)
guy keren <[EMAIL PROTECTED]> wrote:
> i noticed there are various environment variables that allow overriding 
> various settings for the resolver library, but they don't seem to allow 
> enough control to actually override such settings

I'd like to (ab)use your question to general "config" mechanisms
available, and some comparisons:
        Environment variables:
                - Per-process
                - Inherited (so we can make them "global")
                - No change (from outside) after process startup
                - Flat name-space (not scaleable)
        Global config files:
                - Global
                - Change may/may-not affect after process startup
                  (e.g: SIGHUP to daemons etc.)
                - Managed case by case (each with its own rules)
                - Arbitrary types/ammounts of data
        Global config database (e.g: Windows Registry, GNOME GConf):
                - Global
                - Change affect processes even after process startup
                - Centralized management (but luckily, GConf doesn't
                  force you to have all config in one big file)
                - Hierarchical namespace (scaleable)

After looking at the pros/cons of each mechanism, I think it will be
illuminating to present another very old mechanism:
        The X toolkit resource database:
                - Global
                - Hierarchical namespace (scaleable)
                - The key heirarchy permit overrides:
                        - per type of application (window class)
                        - per instance of application (window)
                - Centralized managment (xrdb(1)), but database
                  may be spread at multiple locations (along
                  the XAPPRESPATH and in the Xserver as well)
                - Application may allow changes to occur after
                  startup (instantiation), see editres(1)

Looks like it takes most of the pros from all methods
(except of inheritance -- designed not Unix specific),
without the cons.

Now, isn't it a pitty that this mechanism is dying? Both KDE
and GNOME chose to develop their own (inferior IMHO) mechanisms.

Hopefully later GConf versions (which isn't GNOME/X specific) will
include the missing parts, so we can return to the flexibility
we had in the 80's :-)

----------------------------------------------------------------
Oron Peled                             Voice/Fax: +972-4-8228492
[EMAIL PROTECTED]                  http://www.actcom.co.il/~oron

"A billion flies _can_ be wrong - I'd rather eat lamb chops than shit."
        -- Linus Torvalds on lkml

=================================================================
To unsubscribe, send mail to [EMAIL PROTECTED] with
the word "unsubscribe" in the message body, e.g., run the command
echo unsubscribe | mail [EMAIL PROTECTED]

Reply via email to