Re: is a value writeable?

[Havoc Pennington <hp redhat com>]

Colm Smyth <Colm.Smyth@ireland.sun.com> writes: 
> True, but adding these internal API's makes it easier to write the API's
> below and might facilitate other API's. For example, a search option
> in a gconfedit GUI should be done within the gconfd daemon, not by
> having a client get all keys and values using the GConf public API's.
>

OK, I'm not opposed to them (though until you actually use an internal
API I see little reason to write it).
 
> The gconf_writeable() API has a slightly extended semantic to
> gconf_key_is_overridden(). It checks for overridden keys as
> well as checking if all writeable databases have this key
> marked readonly (using the backend 'writeable' method).
> The effect for a gconf client is identical, so I would
> like to see both checks combined in one API checking writeability.
>

I missed that, that seems sensible. So let's export a gconf_writable()
routine. If we use your idea of a config source map, replacing the
readable/writable methods on the backends, we can even make this
fairly efficient.

Yesterday I noticed that my /usr/dict/words says it's "writable" with
no "e", and I have screwed this up throughout GConf. Hrm. ;-)

Havoc