jackrabbit-oak-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Carsten Ziegeler <cziege...@apache.org>
Subject Re: slow observation tests
Date Wed, 23 Oct 2013 11:40:59 GMT
On the other side, if the observation event arrives way later than the
content change, the event is pretty useless. The more time there is in
between, the higher the possibility that the content is already different
from what the event tells you. I know that this always can happen as events
are async, but the longer the time, the more problematic it gets


2013/10/23 Bertrand Delacretaz <bdelacretaz@apache.org>

> On Wed, Oct 23, 2013 at 11:03 AM, Marcel Reutegger <mreutegg@adobe.com>
> wrote:
> >> ...i recently discussed this with christian keller and he argued
> >> that this is likely to cause troubles in productive systems.
> >
> > hmm, but even jackrabbit does not guarantee event delivery
> > within a certain timeframe....
> FWIW, I had to fix some Sling tests to have them run with Oak, due to
> similar issues where the delivery timing of observation is different
> in Oak.
> One could argue that that's actually a good thing, as it exposes
> incorrect assumptions in the client code - so maybe it's good to lower
> the delay by default, but set it much higher in continuous integration
> testing, to expose those issues.
> -Bertrand

Carsten Ziegeler

  • Unnamed multipart/alternative (inline, None, 0 bytes)
View raw message