Jörn Nettingsmeier wrote:
> Markus Angst wrote:
>> Jörn Nettingsmeier wrote:
>>> but the publication should be *able* to override anything, right?
>> The idea was that a command line property should be able to override even a
>> publication property, but i don't have very strong feelings about this
>> particular issue.
>
> makes sense imho. i hadn't really thought of command line properties,
> but yes, they should be the mostest mightiest.
>
>>> so it ought to be at the beginning if the parsing follows "first one wins".
>> Regarding the point in time the publication properties are read in last (if
>> nobody else finds a better solution) so the order of precedence has to be
>> kept
>> artificially by applying some magic (like leaving alone some properties with
>> particular name prefixes and overwriting others...).
>
> ok, so then forget my remark about parsing order. it's "order of
> precedence" then, which could be accomplished by
> if (property) {
> remove(property);
> add(property,otherSource);
> }
Yes. That's already done for system and publication properties (in
overwriteProperty()) but because the system properties are processed in
initialize() and the publication properties are lazy loaded - well just as i
write this i realize that this could probably be simplified by lazy loading the
system properties, too.
> i don't think prefixes should have an impact on whether a property can
> be overridden. they should just be there to avoid *accidetal* overriding
> when names clash.
I agree. Lazy loading the system properties would eliminate the magic i
mentioned above.
> i'm a little confused about some things:
> can there be a lenya.properties in the core (i.e. under webapp/lenya)?
Yes, there is one under webapp. It is even used, it contains a property of
arguable usefulness called
<property name="pubs.default.author" value="the Lenya community"/>,
which is used when creating a new document. Take the file away and Lenya will
complain. :-)
> can there be one lenya.properties per publication?
Yes, that is the idea of the patch.
> or does that have to be local.lenya.properties?
No. I think one properties file per publication would be enough.
> will template publications be taken into account? if so, we should mimic
> (or better yet, re-use) an aggregating fallback behaviour with
> last-one-wins as described above.
Good point.
> are you doing that already?
Not that i am aware of, but i will have to check.
Thanks!
Markus Angst
---------------------------------------------------------------------
To unsubscribe, e-mail: [EMAIL PROTECTED]
For additional commands, e-mail: [EMAIL PROTECTED]
