GNIP 5 - Abstract OGC:CSW Functionality
David Winslow
Tom Kralidis
________________________________
> Date: Tue, 10 Jan 2012 10:52:18 -0500
> Subject: GNIP 5 - Abstract OGC:CSW Functionality
> From: dwin...@opengeo.org
> To: geono...@opengeo.org
>
> See proposal on the wiki
> at https://github.com/GeoNode/geonode/wiki/GNIP-5----Abstract-OGC%3ACSW-Functionality
>
> This proposal is mostly about a refactoring right? Moving some of the
> custom code we have on top of OWSLib into the main library, like this
> stuff:
> https://github.com/GeoNode/geonode/blob/master/src/GeoNodePy/geonode/maps/views.py#L1344-1412
>
Correct. The work done so far does this (i.e. rather than geonode interacting directly with the CSW XML, push back to owslib). This has resulted in changes to owslib and pycsw, however we can cut releases of both when the time comes.
> Allowing the use of non-GeoNetwork CSW services seems likely to affect
> administration, I'd like to see a bit more in the proposal about how
> administrators will select the backend they need, etc.
>
At this point, src/GeoNodePy/geonode/settings.py has been updated for abstracted CSW settings. The code takes care of subtleties across CSW server implementations (i.e. GN logging in, axis order discrepancies, etc.).
Currently, GeoNode comes with geonetwork as default. We could also bundle pycsw, or put documentation to users on how to configure against a different CSW.
..Tom
Jeffrey Johnson
At a minimum, we need documentation on how to configure GeoNode with
the various backends, including some description of the subtleties of
each and any pitfalls that may be encountered. Im +1 on shipping
GeoNode with pycsw as the default, but there are bound to be varying
opinions on that.
> ..Tom
>
>
Tom Kralidis
> Subject: Re: GNIP 5 - Abstract OGC:CSW Functionality
> From: jjoh...@opengeo.org
> To: tomkr...@hotmail.com
> CC: dwin...@opengeo.org; geono...@opengeo.org
..Tom
Chris Holmes
Jeffrey Johnson
> Switching to pyCSW as the default seems to me like a more major /
> potentially controversial decision than most any other GSIP I've seen go by.
> I think it could make sense to clear out the queue and get a lot of that
> new functionality on before thinking about switching from GeoNetwork as the
> default. I'm all for the work to make things work with any CSW, but think
> we should weigh more strategically what we're doing for the default.
>
> I may be psyched on pyCSW, I just know very little about it, and have never
> seen it in the wild, or even heard anyone talk about it outside of this
> list. So a GNIP on what it provides, and what it doesn't, would be good.
> But please hold off a bit on that one, so we can churn through the ones on
> the queue already.
Agreed. Changing the default csw backend should be another GNIP in and
of itself, and I dont think its necessary that we take it up now.
Continuing with GeoNetwork as the default backend and also providing
documentation on how to configure others (including pycsw and deegree
and potentially others) is the best course of action in the
short-medium term.
Tom Kralidis
Chris: thanks for the info. Sounds good for another GNIP to tackle at a later time.
Date: Wed, 11 Jan 2012 15:58:59 -0500
Subject: Re: GNIP 5 - Abstract OGC:CSW Functionality