|
|
|
I suppose we should cut a quick 1.6.2 release, with just the one fix. It's fine if it's just what's in 1.6.1 plus this fix.
Yeah, 1.6.x already has watermarking in, and I think most of the developer still haven't had a look at it. I would feel better doing an svn copy of 1.6.1 tag into 1.6.2, patch there, and re-release
it would be nice if on deployment Geoserver looked to see if a data dir was installed, and if not provided an option to choose it at the admin screen.
other requests would throw a meaningful error message of course That would require a major change in how GeoServer is started up unfortunately, we'd have to make the Data class an indirection to the actual data class. At the moment Data loads the data dir content during the spring context startup, this basically mean, either the data dir is there where it's supposed to be, or GeoServer won't wire up properly and die.
I agree it would be advisable to do what you say, we have to consider it when we rewrite the config and ui subsystem. Tested a different approach which requires less modifications: we just stick the file hider servlet on the /data/* path and that effectively hides whatever is in the data dir. The file publisher then exposes just what we want.
This approach is better, Jody made me remember that we moved the data dir out of WEB-INF because otherwise under certain web containers it would have been unreachable (was that jboss? can't remember). Integrated in geoserver-trunk #375 (See http://gridlock.openplans.org:8080/hudson/job/geoserver-trunk/375/
|
||||||||||||||||||||||||||||||||||||||||||||
Shall I commit on 1.6.x, people do test it, and then we decide what to do about an eventual re-release? Or we roll out an emergency 1.6.2 by copying what's in the 1.6.1 tag?