Thanks for the link, Robert! did use getItemOrNull too for getParent in
https://issues.apache.org/jira/browse/SLING-8761
(https://github.com/apache/sling-org-apache-sling-jcr-resource/pull/5)

tell me what you think

Le mer. 2 oct. 2019 à 22:51, Robert Munteanu <[email protected]> a écrit :

> On Wed, 2019-10-02 at 18:47 +0200, Nicolas Peltier wrote:
> > I would love this! i'm sure that ca config is not the only one to use
> > getParent() and its null return value though, and wonder if it is
> > "ok" to
> > mask that potential AccessDeniedException to the developer that will
> > make
> > use of it without knowledge of its cost.
>
> What I was speaking of is
>
> https://github.com/apache/sling-org-apache-sling-jcr-resource/commit/f03954d
> . Not sure if it applies to your scenario, probably. If you can see a
> way we can optimise this kind of access, I'd love to hear about it.
>
> Thanks,
> Robert
>
> >
> > Le mar. 1 oct. 2019 à 16:06, Robert Munteanu <[email protected]> a
> > écrit :
> >
> > > Hi,
> > >
> > > On Fri, 2019-09-27 at 17:47 +0200, Nicolas Peltier wrote:
> > > > Hey
> > > >
> > > > realized on applications using caconfig that if you had read
> > > > right
> > > > only on
> > > > /conf/<child>, and not on /conf, each (unsucessfull) lookup ended
> > > > up
> > > > in a
> > > > quiet AccessDeniedException at jackrabbit level, ending up in a
> > > > null
> > > > return
> > > > value.
> > > > It "works", but still under performs (exceptions are expensive at
> > > > that
> > > > intensity). I think that in our context, we can set read rights
> > > > on
> > > > /conf,
> > > > but i wonder if we shouldn't have "conf roots" for ca configs, or
> > > > "root
> > > > depth" whose lookup is done before reaching JCR.
> > > > wdyt?
> > >
> > > I seem to recall that at some point we had a way of accessing
> > > resources
> > > without getting exceptions, just null back in case of problems. I
> > > don't
> > > recall the specifics, but maybe this API can be used for the
> > > caconfig
> > > as well?
> > >
> > > Thanks,
> > > Robert
> > >
> > >
>
>

Reply via email to