On Sat, 2005-07-16 at 02:26 -0700, Mark K. Kim wrote:
> There are some minor issues but it's certainly duable, even for non-*nix
> systems. I think it's prodent to keep it at a discussion stage until some
> of the big-name projects use it. Do you know how much following there is
> for the XDG Base Directory specs?
>
> Some minor issues I'm thinking about:
>
> 1. When parsing the $XDG_*_DIRS environment variables on multiple OSes,
> since the directory separator is different on different OSes, a
> different call would have to be made on different OSes. Some extra
> care may need to be taken (but probably not necessary) when the
> code is called on multi-platform environments.
Worse yet, environment variables just aren't the normal
native mechanism on the various crufty proprietary systems.
Heck, the usage of environment variables may be on the
decline for Linux-like systems. People are starting to
prefer RPC (Corba, XML, etc.) calls and data associated
with the root window.
> 2. If data files exist in old directory, but $XDG* environment variable
> defines a new path, should the old directory be used instead, or
> should the old directory be transferred over, or should the old
> directory be used only for reading old data but the new directory
> be used for writing any new data?
>
> 3. The spec isn't 1.0 yet. Should we wait until 1.0?
>
> 4. The fallback paths aren't backwards compatible. Should the fallback
> paths use the convention that's being used now?
There's also the more fundamental question of usefulness. Who is
going to set this environment variable, and why would they do so?
|
