WIP: add buffers to layers (according to layer configuration). #6

Author: @stirringhalo

https://github.com/MattBlissett
Multithread application — this is done badly, as I don't know the proper
way to do it in Python.

Author: @stirringhalo https://github.com/MattBlissett Multithread application — this is done badly, as I don't know the proper way to do it in Python.

Author: @MattBlissett

Hi! I didn't actually open this pull request, it's very much a WIP (work in "progress") branch on my clone, and a mix (or mess) of buffer, projection, and testing-UI changes.

I used it to produce the tiles you can see on https://tile.gbif.org/ui/, which are in EPSG:4326, EPSG:3575 and EPSG:3031 projection. The code is very rough because

I rarely use Python
I think I was about to go on holiday for a couple of weeks, and left it running during that time. It finished, and I moved on with other work when I returned.
I don't expect to repeat the process very often, and the general web mapping community didn't seem interested in alternative projections, especially these "difficult" ones.

I'd like to revisit it, but it's not likely to be a priority for at least 6-12 months. It's probably best to start again, and just copy bits than you need and make sense, since the projection changes are confusing.

Author: @MattBlissett Hi! I didn't actually open this pull request, it's very much a WIP (work in "progress") branch on my clone, and a mix (or mess) of buffer, projection, and testing-UI changes. I used it to produce the tiles you can see on https://tile.gbif.org/ui/, which are in EPSG:4326, EPSG:3575 and EPSG:3031 projection. The code is very rough because - I rarely use Python - I think I was about to go on holiday for a couple of weeks, and left it running during that time. It finished, and I moved on with other work when I returned. - I don't expect to repeat the process very often, and the general web mapping community didn't seem interested in alternative projections, especially these "difficult" ones. I'd like to revisit it, but it's not likely to be a priority for at least 6-12 months. It's probably best to start again, and just copy bits than you need and make sense, since the projection changes are confusing.

Author: @ianthetechie

@MattBlissett hey man, thanks for the reply! Good to know.

I'm already working on an internal fork of part of the project code anyways and was thinking of working up a PR myself. The lack of buffering in the original was a little concerning to me though when it comes to rasterization. I was looking through this code, basically copying it over into mine and reviewing as I went and that's why I left a review. Sorry I didn't notice that you didn't even open the PR ;)

Cheers,
Ian

Author: @ianthetechie @MattBlissett hey man, thanks for the reply! Good to know. I'm already working on an internal fork of part of the project code anyways and was thinking of working up a PR myself. The lack of buffering in the original was a little concerning to me though when it comes to rasterization. I was looking through this code, basically copying it over into mine and reviewing as I went and that's why I left a review. Sorry I didn't notice that you didn't even open the PR ;) Cheers, Ian

Author: @nyurik

@stirringhalo @ianthetechie @MattBlissett I have used some of the ideas in this and other PRs, and did a full rewrite for MVT generation directly from SQL. The actual tile generation SQL is now part of the openmaptiles-tools repo, because that's where we have all the code generation from the layers. The postserve code has been heavily rewritten too to support the new method - thus greatly simplifying it -- it simply loads an SQL file, runs it to prepare the statement, and then calls gettile(z,x,y) (without any special math) to generate the needed vector tile.

SQL generation code: https://github.com/openmaptiles/openmaptiles-tools/pull/34
new postserve: https://github.com/openmaptiles/postserve/pull/11

Author: @nyurik @stirringhalo @ianthetechie @MattBlissett I have used some of the ideas in this and other PRs, and did a full rewrite for MVT generation directly from SQL. The actual tile generation SQL is now part of the `openmaptiles-tools` repo, because that's where we have all the code generation from the layers. The postserve code has been heavily rewritten too to support the new method - thus greatly simplifying it -- it simply loads an SQL file, runs it to prepare the statement, and then calls `gettile(z,x,y)` (without any special math) to generate the needed vector tile. * SQL generation code: https://github.com/openmaptiles/openmaptiles-tools/pull/34 * new postserve: https://github.com/openmaptiles/postserve/pull/11

Author: @nyurik

@MattBlissett hi, have you seen the new https://github.com/openmaptiles/openmaptiles-tools#generate-sql-code-to-create-mvt-tiles-directly-by-postgis - it generates the needed postgres scripts that you are dynamically creating here.

Author: @nyurik @MattBlissett hi, have you seen the new https://github.com/openmaptiles/openmaptiles-tools#generate-sql-code-to-create-mvt-tiles-directly-by-postgis - it generates the needed postgres scripts that you are dynamically creating here.

Labels Milestones

WIP: add buffers to layers (according to layer configuration). #6

Step 1:

Step 2: