[gis_info] _\|/_ Potentially Risky URL in Email - Click Carefully _\|/_ 3D Map + Model Using Open Source

[John L. Poole]

I used the open source project MapLibre to build a vector tile map 
server (martin <https://github.com/maplibre/martin>) of Open Street Maps 
data for the State of Oregon.  Having a vector tile server allows me the 
opportunity to create a 3 dimensional area (with some limitations) and 
display a model in it and have the model move about.  Everything is nice 
and crisp and does not suffer the distortions bit mapped images can 
cause.  I was experimenting with QGIS and the ThreeJS plugin and decided 
I had to have a vector tile server if I was going to be serious about 3D 
modeling with maps.  I was not inclined to subscribe to a commercial 
service nor give up my privacy.  So I built my own tile server.   While 
I'm currently working off of a SQLite database (~430 MB), the martin 
server (built in rust) ties in with PostGIS -- I haven't explored that 
as yet, it was enough to get the tile server with a tile set of the 
State of Oregon working on a TLS [fna SSL] server.   I have some working 
knowledge of databases given my previous employment for 24 years at the 
worlds largest database company.

I started off with the Open Map Tiles 
<https://github.com/openmaptiles/openmaptiles>project and then modified 
their broken style sheet (with the help of the MapLibre Maputnik 
<https://github.com/maplibre/maputnik>project which also had problems I 
had to overcome) and got a tile server to work with Apache, I prefer 
Apache over Node.  (I'm a seasoned Gentoo Linux enthusiast and Node, 
like PHP, just gives me the willies.) Then I built the martin server in 
rust -- that was a piece of cake, and then the trick was to tie in the 
martin server with Apache and overcome CORS and proxy issues -- that 
took a tremendous amount of time.  A take-away: the paradigm is 
extremely fragile/brittle, one setting off and nothing renders.

I could not have gotten to where I am without the help of ChatGPT -- 
it's like having a junior software assistant to figure the mundane tasks 
that would take me a lot of time, e.g. creating a model simulation based 
on ads-b flight data so that it mimics in real time the actual flight in 
a 3D map environment.  Example: I said I had a model in a 3D map 
environment and wanted to feed it my captured ads-b data to make it 
simulate in real time the actual flight path.  Boom: proposed JavaScript 
to accomplish it.  I wanted a circular data set of artificial ads-b data 
to do testing with:  centered on my property on Court Street, 1/2 mil 
radius, sinusoidal elevation from 100 to 500 and back again in one 
cycle, intervals of coordinates every 1 second, or every 50 
milliseconds.  Boom: a Python script that creates a data set. Then I 
thought it might be nice to generate the set on the fly within 
JavaScript so a user could create a model on a fly and then watch the 
aircraft fly the route.  Boom: ChatGPT converted the Python script to a 
JavaScript returning to me the array of data. I got a lot of exercise as 
I went up the staircase from my basement office to tell my wife what 
miracle ChatGPT had performed in software design.  (AI can absolutely 
fail in historical data by "hallucinating" ~= lying.)  I'm ready to 
testify that given the right handling (and understanding of its 
limitations), AI will revolutionize software development.

    https://salemdata.net/johnpress/?p=170

Out of pocket costs: $0 (though I've now decided to obtain a premium 
account with ChatGPT for $20/month because I'm convinced it has saved 
10s, if not 100s of hours).

My time -- only The Lord knows (a couple of months?).

I know everyone here are acolytes of ESRI, but some of us have to work 
on a shoestring on our own and believe in open source solutions.  I felt 
it appropriate to share this story and possibly inspire some considering 
open source alternatives.

Cheers,

John

-- 
Email Rider

John Laurence Poole
1566 Court ST NE
Salem OR 97301-4241
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://omls.oregon.gov/pipermail/gis_info/attachments/20240730/407df539/attachment.html>