Stewart Stremler wrote:
> begin  quoting [EMAIL PROTECTED] as of Fri, Sep 22, 2006 at 02:29:51PM -0700:
>> Like me you probably have multiple Linux boxes & multiple home directories
>> you work on.
>>
>> If so you probably have scripts and programs you copy to all your
>> home directories.
>  
> Actually, I stopped, as I discovered that each new system is an
> opportunity to experiment (and to prune out the cruft that isn't
> needed).  And because the scripts I'd haul around from system to
> system were getting bloated with irrelevant cruft.
> 
> And by starting anew on a new machine, I manage to avoid getting
> too caught up in the "oh, look at all the bizarre customizations
> I can do!" bandwagon.
> 
>> If you make changes to files you may end up with 'version creep'
>> as all your home directories slowly diverge.
>>
>> Any advice how to manage this complexity?
> 
> Put it all under version-control.
> 
> If it's all on a local network, use an NFS-mounted (or equivalent)
> home directory from a central server.

I have some similar remarks. If every distro/install wants to customize
my home directory differently, I've decided not to fight. Instead I put
my _real_ personal work in another subdir (jbase <as in all-my..>) off
of /home (but it could be a subdir of ~, I suppose). Now I don't have to
look at those [EMAIL PROTECTED]&* dotfiles/dirs either.

All important stuff in that tree (almost everything, it turns out) is
under source control management (svn, multiple 'project' repositories
under /srv/svn-repos).

That leaves things like browser bookmarks and email archives (in default
installs) still under some ~/.dotsomething/. I haven't worried too much
about that yet, but I suppose a sensible plan would be to customize
and/or symlink those things into jbase files/dirs and get them into svn
as well.

My concern hasn't been for multiple machines as much as for migration
upon upgrade, but the above remarks probably apply to both situations.

Now for multiple machines, I personally would probably view the
challenge as replication rather than realtime consistency -- since I'm
too cheap to pay for powering a separate fileserver <heh>. The
segregation of my real stuff from OS/disto stuff seems like it should
make rsync recipes a lot simpler.

BTW, you may have noticed the use of /srv for my repositories. I like
the FHS invention of /srv as a home for server-specific data. I (try to)
always locate or symlink all server-data there (mysql, apache, etc).

I haven't gotten as close as treed to having everything unique about my
system under source control .. but I'm getting better.

Regards,
..jim


-- 
[email protected]
http://www.kernel-panic.org/cgi-bin/mailman/listinfo/kplug-list

Reply via email to