[Community] CSW Review (Re: [SVN commit] (gispy) r1289 -OWSLib/trunk/owslib)

[Kralidis,Tom [Ontario]]

> -----Original Message-----
> From: Dominic Lowe [mailto:dominic.lowe at stfc.ac.uk] 
> Sent: Monday, 08 June 2009 05:03
> To: community at lists.gispython.org
> Cc: Kralidis,Tom [Ontario]
> Subject: Re: [Community] CSW Review (Re: [SVN commit] (gispy) 
> r1289 -OWSLib/trunk/owslib)
> 
> Hi Tom,
> 
> > Here's where I'm falling down: if lxml proper is not used, the 
> > namespace support breaks, especially when constructing the 
> XML message 
> > required for HTTP POST to a CSW.
> 
> I'm not quite sure what the issue is, but ElementTree 
> contains an internal namespace map which you can add 
> namespaces to - thus avoiding 'ns1', 'ns2' 
> type output (if that's the problem here).
> 
> See "patch_well_known_namespaces" here:
> http://trac.gispython.org/lab/browser/OWSLib/trunk/owslib/etree.py
> 

Thanks (to Sean as well for this recommendation).

I've made an update to trunk.  Review and comment would be appreciated.

> And also the section entitled "Element Tree Representation" here:
> http://effbot.org/zone/element-namespaces.htm
> 
> hope that helps,
> 
> Dom
> 
> On Friday 05 June 2009 23:16:08 Kralidis,Tom [Ontario] wrote:
> > >Tom,
> > >
> > >I'd like to see classes and methods similar to the ones in the 
> > >existing modules (wms.py etc), providing the same interfaces. For 
> > >example
> > >
> > >   class CatalogService:
> > >       """instead of request"""
> > >
> > >       def getrecord(self):
> > >           """instead of GetCapabilities"""
> >
> > Done.
> >
> > >This is also more in line with Python PEP 8, capitalized 
> class names, 
> > >lowercase method names. Any service should have a "capabilities"
> > >property. See the wms.py, wfs.py, and wcs.py modules.
> > >
> > >Importing etree from owslib.etree tries lxml and then 
> falls back on 
> > >elementtree or xml.etree, so do that instead of depending 
> only on lxml.
> >
> > It looks like lxml is tested last, no?
> >
> > Here's where I'm falling down: if lxml proper is not used, the 
> > namespace support breaks, especially when constructing the 
> XML message 
> > required for HTTP POST to a CSW.
> >
> > Using elementtree or xml.etree, I haven't found a way (or 
> examples) to 
> > construct valid XML requests.  Here's an example:
> >
> > http://devgeo.cciw.ca/cat/GeneralQuery1.xml
> >
> > Any suggestions?
> >
> > >Pedantic me, I can not abide the name "fetch" for a method 
> that makes 
> > >an HTTP POST request ;) GET is the HTTP method with fetch 
> semantics.
> > >Call it "_doRequest" or something. Probably shouldn't be 
> part of the 
> > >public API.
> >
> > Done.
> >
> > >Lastly, let's get some tests in there soon.
> >
> > In the same way as wms/wfs/sos, right?  Putting in 
> docstrings right in 
> > the classes might bloat the code (responses can be big).
> >
> > >I hope you don't mind the review. I just wanted to point out some 
> > >little things.
> >
> > Not at all!  Thanks alot for your advice and irc help!  
> Much appreciated.
> >
> > I haven't committed the updates yet, pending discussion on 
> the lxml issue.
> >
> > ..Tom
> > _______________________________________________
> > Community mailing list
> > Community at lists.gispython.org
> > http://lists.gispython.org/mailman/listinfo/community
> 
> 
>