Feature Editing

[theduckylittle]

Implement WFS-T layers.

• Geometry editing
• Attribute editing

We'll use TinyOWS as the example WFS-T server but GeoServer should be tested as well.

[brentfraser]

Where are we on this?

[theduckylittle]

Sort in bullet points:

• TInyOWS is essentially dead.
• GeoServer would be our next best target.
• WFS-3.0 changes how this is all done.
• There needs to be some substantial review of the vector editing tools in general.

The blockers at this point are:

• We don't have any real sponsorship for the feature.
• We don't necessarily have a solid server-side target or spec folks would like to see.

I'm in favor of moving this forward but want to make sure it's a solution that fits how people are putting things together in 2019.

[tchaddad]

I could be wrong, but I think there is quite some time before WFS 3.0 is available, so any solution in the shorter term may need to be based on the existing standard.

Edit: here is the URL for the location where it (now renamed OGC-API) is being developed:

https://github.com/opengeospatial/ogcapi-features

[klassenjs]

A lot of the base code needed for editing features is there. We can draw/edit features on vector layers. This can be done in the original layer or the features can be selected from a source layer and brought into a temporary layer (generally the results layer, but could be others). The user can upload/download features from a vector layer to/from GeoMoose as GeoJSON or KML. I'm not sure if we have attribute or style editing in GeoMoose 3.x yet.

The part that is currently left as an exercise for the reader is getting this back into a database. This could be done with a custom service (JS in GeoMoose) and something on the server to go between HTTP and the database. Ideally the HTTP part would be some open standard API with a FOSS implementation and growing mindshare, but I'm not sure what that would be at the moment.

[brentfraser]

Hmm. Reminds me of GM 1.6 (not that it's a bad thing)...

[brentfraser]

One thing I liked about the original vector editing was the performance when dealing with lots of vectors. The layer with all the features was rendered with WMS, the user identified/selected a feature (or multiple features?) for editing, the geometry of only that feature was retrieved and presented for editing. Good stuff.

[tchaddad]

Is TinyOWS non-functional? If it still works, maybe incorporating it as a placeholder until the new standard is operational is the best path...

[theduckylittle]

My worry is that TinyOWS is not-maintained. The current OGC spec only specifies the way to read features and not the way to write or update them.

I think tagging this to 4.x is likely the best way to go. I don't want this to fall off the list but it's definitely a feature that could use some sponsorship.

[jmckenna]

For the record, some much-needed recent activity for TinyOWS: MapServer/tinyows#94

Also, please keep in mind HTTPS support, in your GeoMoose-WFS-T testing (MapServer/tinyows#93).

[brentfraser]

So the current TinyOWS (including PR 94) will not work with a https request?

[jmckenna]

I think it needs re-testing, and then update the ticket with test results at MapServer/tinyows#93

[brentfraser]

see PR #620