Re: Structure in $HOME

On 12 Jan 2003, Andrew Sobala wrote:

> I might want to back up all of my preferences, for backup reasons,
> without backing up the whole galeon cache. For this reason, it would be
> easier to have a single preferences directory to back up, with
> application-specific data in separate directories after that.
> So preferences could be stored centrally, and I could additionally back
> up evolution (all my emails) but not certain other programs (eg.
> galeon). Which isn't so far from what we have at the moment.
> I don't agree that it should be unhidden though; I hate the way
> evolution stores its data in an unhidden directory. A user shouldn't
> need to access program data files 99.999% of the time - backups, that's
> pretty much it. Installing fonts and themes should have a reasonable
> GUI.

I agree with Andrew here. The current situation is more or less that each
application and framework store their data below their own directory
first, and then divide it in parts. However, looking at the root structure
(below /) it is the opposite: /etc, /usr/share etc. When it comes to
backup, I am mostly interested in backing up data that comes from a
conscious choice of mine. Cached data or lock files that are created by
the applications themselves is of little use to back up. I want to be able
to point to a directory and say: here are my preferences, and only that.
In the same way I want to know that here are my bookmarks, here is my
mail, here is my address book. These are not preferences but my personal
data. And my cache, cookies and lockfiles I don't care about.

If each application keeps its own data in its own directory it is harder
to integrate applications with each other. Sharing becomes dependent
on what applications are installed, and each application will continue to
rule their own little world. I would like a directory structure that eases
sharing of resources, and sharing is improved if we think in terms of data
rather than applications.

I think that "ordinary users" will not back up files by copying
them directly. They would rather user tools for this. What is important is
to make a directory structure that makes sense for developers. Then it
will be easier to write backup tools if the directory structure is
arranged in a way that files with different properties are separated in
different directory branches.

Finally, while I also think that preferences and application data should
be hidden, at least at the top level, I think that discussion is not
really important at this point. Instead we should find a directory
structure that makes sense. What we should try to do is to identify what
different kinds of preferences that exist, what kinds of temporary,
"discardable" data that exist, and what kinds of resources (like
bookmarks, contact lists) that exist, and build a structure that makes
sense for that.


[Date Prev][Date Next]   [Thread Prev][Thread Next]   [Thread Index] [Date Index] [Author Index]