jackrabbit-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Thomas Mueller <muel...@adobe.com>
Subject Re: [jr3 microkernel] Change log consolidation
Date Mon, 06 Feb 2012 14:46:48 GMT

>> (a) unnecessarily complicated
>i don't agree, why would that be complicated?
>when using a content-addressable model diff'ing get's even
>extremely efficient.

Well, "extremely efficient" is relative. "Diffing" is definitely more
complicated than, well, not diffing. As an example, I know of no database
that tries to re-construct the SQL statements from the diff of the stored

>i partially agree. however, the current jackrabbit implementation
>creates events based on diff'ing states and IMO it works pretty
>well in practice. i am not aware of real world uses cases affected.

We know some cases that don't work (reorder) but I agree they are not that
important. But I believe the current solution is rather complicated.

>personally i don't see the practical value in representing multiple
>operations within a single save rather than consolidating them.

My approach is: keep the events as called by the application. I think
that's rather simple. The other approach is: from the events change the
state and then try to re-construct the events from there. That sounds
rather complicated and unnecessary to me, because originally you *had* the
events. Why did you throw them away?


View raw message