Martin Desruisseaux ha scritto:
> Jody Garnett a écrit :
>> I would like more options rather than less; Derby has been rolled into
>> the standard Java 6 install so it may be applicable to more users.  I
>> would *love* to tell applet developers to use Derby (enabling us to
>> retire the epsg-wkt plugin).
> 
> I have a preference for Derby as well because it is supported by Apache and 
> Sun
> (maybe IBM too, not sure), is included in JDK 6 and would eliminate the need 
> for
> writting in a temporary directory. If it is slower than H2 because it performs
> lock at row level instead of table level, it may be a good reason (maybe not 
> for
> EPSG factory alone since it is read-only, but if it used on a wider basis).
> 
> However if the Geoserver community choose H2, we should probably support a 
> EPSG
> factory for H2 in Geotools code base.

Well, it's not that we really "chose" H2 because there is nothing 
sizeable going on. It just seems the most likely candidate for an 
embedded spatial db because it's apparently the fastest of the back
and does support (in a limited way) multidimensional indexes.

It was also the easiest alternative to try out, since it's API 
compatible with HSQL... I really had to make really thin changes to the
current HSQL code to make it work against H2.
The same won't probably apply to Derby.

I also would like to notice another thing. The current HSQL provider
uses more than 4MB or memory to keep a cache of the tables in the
DB (that's what the profiler reports, HSQL is the single biggest stable
memory user in GeoServer).
I find this wasted memory annoying, since most of the GeoServer
installations would in fact just use 3-10 CRS, but in an Applet
that would be a real killer.
Derby has the ability to work out directly of the classpath, but if
I'm not mistaken, this means it has to load in memory the whole content
of the db, since random access is not allowed in the classpath, or
alternatively has to unpack the db contents on a temporary storage.
The first option would use tons of memory, the second would not be
any better than the current one.
That said, having some tests using Derby too would be real cool
(we can talk a lot, but testing provides a definitive answer) :)

Cheers
Andrea

-------------------------------------------------------------------------
This SF.net email is sponsored by: Splunk Inc.
Still grepping through log files to find problems?  Stop.
Now Search log events and configuration files using AJAX and a browser.
Download your FREE copy of Splunk now >>  http://get.splunk.com/
_______________________________________________
Geotools-devel mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/geotools-devel

Reply via email to