Closed GoogleCodeExporter closed 9 years ago
[deleted comment]
We put this to Adam a month ago for any clarification he can bring but I have
not been able to get a specific response. We need to address this as this
function is important to address Issue such as #76 (dependent on this issue and
posted 30/8/2011?) and any Area Report based on polygon classes. Suggest
expertise to address this is required by at least one in the team?
Original comment by leebel...@gmail.com
on 11 Sep 2013 at 12:11
Email sent to Adam and Lee on 26/11/2013:
Thanks Adam, okay I'll send through some questions via email. Let me know if
you think it would be better to Skype about any particular points/issues.
When we last spoke before your internet dropped out, we were talking about the
use of layers loaded using the GridClassBuilder (e.g. Dynamic Land Cover).
I mentioned that I had fixed (fix not yet deployed) the web service that
returns the wkt for layer classes and objects to correctly handle these layers.
Problems still remain in the spatial portal however. In a number of cases the
"placeholder" wkt for classes (e.g. "ENVELOPE(cl918:15)") is being used
verbatim when running a number of spatial portal functions which is causing
problems. For example the "ENVELOPE" wkt is sent to the distributions service
when trying to add species occurrences to the map restricted to an area.
There is the issue with the area report but we have discussed that previously
and I will fix the problem that you identified for me.
An additional problem is that most of the wkt classes generated by the
GridClassesBuilder are invalid geometries due to self-intersecting polygons etc.
Questions (Lee - can you please provide your input on these where applicable):
1. You mentioned that in some cases, bounding boxes for GridClassBuilder
generated classes should be used with spatial portal functions instead of the
geometries themselves due to their complexity. Can you (and possibly Lee?) help
me make a list of what functions should use bounding boxes/what functions
should use the actual geometries?
2. The "ENVELOPE" wkts are not being intercepted and substituted. Is there an
area of the spatial portal code that you would suggest examining for problems
in this regard? You mentioned that these geometries were being handled
correctly at some point so it sounds like some defects have been introduced.
3. You mentioned that there may be issues with using some of the
GridClassBuilder classes in the spatial portal due to their large size. Did you
and Lee have a strategy for how to deal with these large geometries, or should
we simply fail gracefully with an appropriate error message?
4. I was thinking of adding an additional step in the GridClassBuilder
ingestion process to fix the invalid geometries, e.g. using postgis functions.
Do you have any thoughts on this?
Thanks for your help I really appreciate it.
Cheers,
Chris
Original comment by chris.fl...@gmail.com
on 23 Dec 2013 at 12:21
Notes on the issue with the Area Report:
Contextual layers loaded with using the GridClassBuilder tool have two fields
table entries, with types marked "a" and "b". "a" is for entire classes and "b"
is for individual polygons.
In the case of the dynamic and cover layer these entries have ids cl918 ("b"
type) and cl1918 ("a" type). The request to the biocache is specifying cl918
which only returns a small number of records. If this request is changed to use
cl1918, it returns what appears to be the correct number of records.
The requests being made to the biocache are using
Original comment by chris.fl...@gmail.com
on 23 Dec 2013 at 1:11
Both layers are type 'a' - they contain multiple polygons with the same
class. If anyone does a 'Add to map | Areas | Gazetteer polygon' on
either of these two layers, selecting a particular polygon should select
ALL polygons of that class.
Original comment by leebel...@gmail.com
on 23 Dec 2013 at 3:47
My previous comment got cut off.
As indicated by Lee, the biocache requests should specify the "a" type field
table entries.
Original comment by chris.fl...@gmail.com
on 23 Dec 2013 at 4:05
As far as I can tell, this seems to be fixed.
Original comment by leebel...@gmail.com
on 5 Jan 2014 at 11:39
Original issue reported on code.google.com by
moyesyside
on 8 Aug 2013 at 12:33