[mapserver-dev] RFC 119 - Vector Tiles...

[Lime, Steve D (MNIT)]

Hi Even! Thanks for the review amd suggestions. I will make the suggested changes... I'll need to look at the WFS code to comment more. That was Thomas' original approach, handle as a qiure

Regarding mime-type. It's not specified in the earlier spec that Thomas worked from (1.0.1 I believe) so I imagine that's where the current value comes from. The 2.1 spec says "should" for mime-type, not "must" so I think there is at least some room for interpretation. I see GeoServer is using "application/x-protobuf; type=mapbox-vector" - somewhere in the middle. Thoughts?

Yes, I was planning on adding tests, I have a bunch developed during development. I would appreciate thoughts on the scope though...

Steve

From: Even Rouault [mailto:even.rouault at spatialys.com]
Sent: Friday, July 21, 2017 4:06 PM
To: mapserver-dev at lists.osgeo.org
Cc: Lime, Steve D (MNIT) <steve.lime at state.mn.us>
Subject: Re: [mapserver-dev] RFC 119 - Vector Tiles...


Hi Steve,



> Call for review please. I need to add the syntax changes for MapCache but

> that's about it. Associated pull requests are:

>

>

> - https://github.com/mapserver/mapserver/pull/5376



* "the implementation introduces a new dependency on protobuf/protobuf-c". Perhaps add "optional" dependency to make things clearer.



* "WMS_SRS metadata (layer and server) should include epsg:3857 and epsg:900913" : From what I see in the code, the implementation is CRS agnostic. Might be good to know for people who would like to do non conventional uses of VT.



* Perhaps the RFC text could mention the EDGE_BUFFER (default value: 10) specific format option so that it isn't forgotten for the documentation. Same thing for the EXTENT (default value: 4096) format option.



* I see mapwfs.c is also impacted. Should probably be mentionned in the RFC text as well. So I see that the extent is enlarged when the BBOX WFS parameter is specified (which makes sense of course). From what I remember, when a more complex filter involves both an attribute and BBOX filter, that doesn't go through that code path, so other changes might be needed else where to also do BBOX extension. But anyway WFS and MVT are likely not commonly used together, so such cases might be dealt later as enhancements if the need arises.



* Mimetype. I see you use a generic "application/x-protobuf" but

https://github.com/mapbox/vector-tile-spec/tree/master/2.1 mandates "application/vnd.mapbox-vector-tile"



* Do you plan to add new test cases in msautotest ?



* Final remark: great work from you and Thomas!



Even





--

Spatialys - Geospatial professional services

http://www.spatialys.com
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.osgeo.org/pipermail/mapserver-dev/attachments/20170725/c03afee9/attachment-0001.html>