Hi,
[EMAIL PROTECTED] schrieb:
If I understand correctly, this functionality does not help Jackrabbit
(makes it more complex), but would allow other projects to store their
class files inside Jackrabbit.
So when Lenya uses Jackrabbit:
1. Jackrabbit loads.
2. Lenya's classes are extracted from Jackrabbit and load.
3. Lenya loads.
Exactly so.
For a performance hit during startup, we gain the ability to store the
entire project within the repository. That could be good, because we
want everything treated as content. The concern is whether this
damages the ability to replace Jackrabbit with another JCR
implementation. If it requires special functions outside JCR (rather
than built on JCR), we could not use it.
The only Jackrabbit functionality required up to now, is that (mixin)
node types may need to be configured (left out by intent from JSR 170) -
and of course that a series of optional JCR node types like nt:file and
nt:folder be present. Apart from that, only JCR standard API is used. If
you can solve defining the node type(s) outside the classloader and
extension framework code by repository specific functionality, you can
use any standards complying implementation you like.
Regards
Felix