It doesn't break anything, it's just that pretty much everything relies on stuff that's in gxp (commonly gxp.util.dispatch()), which relies on stuff in geoext/ which relies on stuff in openlayers and extjs. So even for the AME page we need the full complement of JS libraries.
Basically it will take a rewrite of the "non-spatial" widgets to make them really non-spatial. -d On 06/28/2010 04:21 PM, Andreas Hocevar wrote: > Hey David, > > thanks for digging into this! And no objections against pushing what you > already have into the main repo. > > If this breaks the AME upload page, I'll be glad to help fixing it. Just let > me know. > > -Andreas. > > On Jun 28, 2010, at 21:36 , David Winslow wrote: > > >> On 06/25/2010 12:54 PM, David Winslow wrote: >> >>> On 06/25/2010 09:50 AM, Andreas Hocevar wrote: >>> >>> >>>> I think there should also be a template for ux. Right now the gxp template >>>> loads gxp and ux resources. In the case of AME upload, only ux is >>>> required, but not gxp >>>> >>>> >>> I broke things up this way, but the AME index page still fails to load >>> its Ext grid if I omit gxp. I guess we will need to rearrange the build? >>> >>> I have updated my repo with the stuff I did so far (the embedded map >>> viewers make a noticable difference even locally, I guess loading Ext >>> once instead of twice is a big deal :) ) in case someone else wants to >>> take a crack at it. >>> >>> Here is the current breakdown: >>> http://github.com/dwins/geonode/tree/fewer-includes/src/GeoNodePy/geonode/templates/geonode/ >>> maybe you had something else in mind? >>> >>> -- >>> David Winslow >>> OpenGeo - http://opengeo.org/ >>> >>> >> I dug into this a bit today. I think that our dependencies are >> currently a bit too tangled for me to just peel out the non-spatial >> components; even the AME file browser uses pieces of GXP. >> >> So, any objections to just pushing what I have to the main repo? We can >> revisit the refactoring later, and maybe bring up some alternative >> tooling options. >> >> -d >> >
