Hi!
Am 13.01.2008 um 15:53 schrieb Jukka Zitting:
[1] http://jackrabbit.apache.org/dev/ngp.html
A few questions came to my mind when reading the NGP proposal:
1) When using a full revision based storage it might be good to look
at the SVN implementation. In fact, for the performance it is
important to have the most-requested nodes and property versions
readily available without long parsing and merging algorithms. This is
typically the most current version - in SVN it's the HEAD and this is
the one that is stored in full text, whereas for older revisions SVN
stores the diffs: to get an older revision, the SVN server takes the
HEAD version and incrementally applies the back-diffs up to the
request version.
2) What about SPI? How will the SPI interface and the NGP relate? Is a
rewritten Jackrabbit supposed to implement the SPI interface only (And
the client part will always be jcr2spi)?
3) Performance for remote storage: my experience showed that the best
performance with a fine-granular API like JCR can only be reached if
the user is able to query the entire set of elements he wants to
operate on in the beginning. Eg. having a good query language (better
than XPath, more like XQuery) that allows the implementation to fetch
the requested subtree(s) at once, so that the subsequent API calls to
individual nodes and properties no longer need extra network
connections. This depends heavily on the way the user uses the API,
but at least he should be able to do so. NGP sounds like something
that should incorporate such an improvement.
Alex
--
Alexander Klimetschek
[EMAIL PROTECTED]
