[mapserver-dev] Vector Tile Production...

Lime, Steve D (MNIT) Steve.Lime at state.mn.us
Mon Sep 14 07:59:53 PDT 2015


Agree with Even - big thumbs up and a great surprise for a Monday. Is this WFS only or could native MapServer query methods also push output through this driver? I'm not quite sure why we'd want to do that yet but much as single tile WMS or MapServer mode=map is useful so might be single tile vector maps. 

I agree w/Even on potentially somehow supplying info necessary to compute scales. I've used that as a trick for doing mode=scalebar sorts of things in the - you set the URL up to represent a corresponding map rendering.

Steve

-----Original Message-----
From: mapserver-dev-bounces at lists.osgeo.org [mailto:mapserver-dev-bounces at lists.osgeo.org] On Behalf Of thomas bonfort
Sent: Monday, September 14, 2015 7:10 AM
Cc: mapserver-dev at lists.osgeo.org
Subject: Re: [mapserver-dev] Vector Tile Production...

Hi all,

I have  a mostly functional and fully untested (on the client side)
branch that implements vector tile rendering for mapserver:
https://github.com/tbonfort/mapserver/tree/vector-tiles . It relies on
mapserver's WFS implementation, so we'll need to also adapt mapcache
to handle these WFS queries in order to present  x/y/z style URLS so
it becomes actually usable.

Given that we're using WFS, we actually don't care about projections,
you can request a vector tile for any extent/projection combo,
although that isn't going to be supported by any client (yet?). I did
have to fiddle around with handling scale computations as there's no
concept of scale in traditional getfeature queries
(https://github.com/tbonfort/mapserver/commit/50c3a44ddd3c182ca0e1a16a6e5a1c5c20675768)
=> Steve or Even could you have a glance?

I could use some help with testing this out, especially on the client
side. Here's how you can get a tile:

* add an outputformat:
OUTPUTFOMAT
  NAME "mvt"
  DRIVER "mvt"
END

* add the name of the outputformat to wfs_getfeature_formatlist
metadata entry of the layers that are allowed to be returned as vector
tiles
METADATA
  "wfs_featureid"     "UNIQUE_KEY"
    "wfs_getfeature_formatlist" "mvt"
    "gml_include_items" "NAME,UNIQUE_KEY,CAPITAL,POP_RANGE"
    "gml_POP_RANGE_alias" "POP_CLASS"
END
(you can use traditional WFS keys to filter attributes, remap
attribute names,  etc...)

* run with map=/path/to/vector_tiles.map&map.size=256+256&SERVICE=WFS&VERSION=1.1.0&REQUEST=GetFeature&TYPENAME=popplace,road&OUTPUTFORMAT=mvt&srsName=epsg:3857&bbox=-20000000,-20000000,20000000,20000000,epsg:3857"

Best regards,
Thomas


On 8 September 2015 at 22:48, Basques, Bob (CI-StPaul)
<bob.basques at ci.stpaul.mn.us> wrote:
> All,
>
> I can come out on the vote for projection support right now.  As I think
> about it, it seems like it might be more work to NOT include projections, at
> least based on the MapServer template outputting.
>
> I would also suspect that this type of feature description would quickly
> start a conversation about 3D (or 2.5D at the very least) types of
> constructs.
>
> bobb
>
>
> On Sep 8, 2015, at 3:20 PM, Rahkonen Jukka (MML)
> <jukka.rahkonen at maanmittauslaitos.fi> wrote:
>
> Hi,
>
> There is a blog post from Stefan Henneberger
> http://www.salzburgresearch.at/blog/geoserver-rendering-binary-vector-tiles/
> that gives some background.
> If the aim is simply to serve vectors in web Mercator to be rendered on the
> client side then probably the MVT way would be fine. Perhaps with some
> modified GetStyles request that would return styles as Style JSON could be a
> good pair with it or what do you think?
>
> I can imagine that very soon there will be a wish to use vector tiles for
> data transfer instead of WFS. It may introduce demands for having support
> for projections, more geometry types, schema,  and need to decide what to do
> with features which are crossing tile boundaries and other complicated
> things like how to download only tiles which are changed since last access.
> A common OSGeo specification could be nice to have at that stage, I do not
> believe that such things bother MapBox so much.
>
>
> -Jukka Rahkonen-
>
>
>
> Lime, Steve D wrote:
>
> The fact ESRI is following the MapBox spec instead of creating their own
> lends a great deal of credibility to the MapBox work. I’ve not run across
> any competing specs – has anyone else? I would assume both OL3 and Leaflet
> would be target client environments. --Steve
>
> From: Tamas Szekeres [mailto:szekerest at gmail.com]
> Sent: Tuesday, September 08, 2015 1:54 PM
> To: Lime, Steve D (MNIT) <Steve.Lime at state.mn.us>
> Cc: thomas bonfort <thomas.bonfort at gmail.com>; mapserver-dev at lists.osgeo.org
> Subject: Re: [mapserver-dev] Vector Tile Production...
>
> Do we already have a standard to follow when implementing this? As far as I
> know ESRI followed the mapbox vector tile specbut that is bound to a
> specific company. What would be the client side application we are targeting
> to consume the vector tiles?
>
> Best regards,
>
> Tamas
>
>
> 2015-09-08 18:27 GMT+02:00 Lime, Steve D (MNIT) <Steve.Lime at state.mn.us>:
>
> Should we start work on an RFC then? I can get a skeletal one in place but
> this enhancement should encompass both MapServer and MapCache at the same
> time IMHO...
>
> Steve
>
> -----Original Message-----
> From: thomas bonfort [mailto:thomas.bonfort at gmail.com]
> Sent: Monday, September 07, 2015 12:51 PM
> To: Lime, Steve D (MNIT) <Steve.Lime at state.mn.us>
> Cc: mapserver-dev at lists.osgeo.org
> Subject: Re: [mapserver-dev] Vector Tile Production...
>
> That's definitely something we should be looking into I'd say.
>
> --
> thomas
>
> On 27 July 2015 at 20:45, Lime, Steve D (MNIT) <Steve.Lime at state.mn.us>
> wrote:
>> With 7.0 out the door I’m wondering if makes sense to look forward at all
>> to
>> being able to produce vector tiles with the MapServer/MapCache stack. It’s
>> been a hot topic with our GIS staff with ESRI announcing some preview
>> capabilities with vector tiles tied to the MapBox spec. --Steve
>>
>>
>> _______________________________________________
>> mapserver-dev mailing list
>> mapserver-dev at lists.osgeo.org
>> http://lists.osgeo.org/mailman/listinfo/mapserver-dev
> _______________________________________________
> mapserver-dev mailing list
> mapserver-dev at lists.osgeo.org
> http://lists.osgeo.org/mailman/listinfo/mapserver-dev
>
>
> _______________________________________________
> mapserver-dev mailing list
> mapserver-dev at lists.osgeo.org
> http://lists.osgeo.org/mailman/listinfo/mapserver-dev
>
>
>
> _______________________________________________
> mapserver-dev mailing list
> mapserver-dev at lists.osgeo.org
> http://lists.osgeo.org/mailman/listinfo/mapserver-dev
_______________________________________________
mapserver-dev mailing list
mapserver-dev at lists.osgeo.org
http://lists.osgeo.org/mailman/listinfo/mapserver-dev


More information about the mapserver-dev mailing list