_______ __ _______
| | |.---.-..----.| |--..-----..----. | | |.-----..--.--.--..-----.
| || _ || __|| < | -__|| _| | || -__|| | | ||__ --|
|___|___||___._||____||__|__||_____||__| |__|____||_____||________||_____|
on Gopher (inofficial)
HTML Visit Hacker News on the Web
COMMENT PAGE FOR:
HTML JupyterGIS breaks through to the next level
eric-modem wrote 1 day ago:
I'm beyond excited to see this as a refugee from esri's products. The
only reason I still use anything esri is due to their layout tools.
Nightmarish as they are, it's still the best way to print or export a
map to pdf. I'd like to see some advances in that area though I would
bet money that Jupyter's architecture isn't adequate for those use
cases.
boxerab wrote 1 day ago:
Looks nice. Hopefully this is one more geospatial jenga piece removed
from the current
dominance of buggy, ancient closed source software like Google Earth
and ESRI.
driggs wrote 1 day ago:
> One of the most significant updates is a new browser-based processing
toolbox powered by a WebAssembly (WASM) build of the Geospatial Data
Abstraction Library (GDAL).
> Available tools include:
> Buffer, Convex Hull, Dissolve, Bounding Boxes, Centroid, Concave Hull
Why would they want to calculate these from WASM in the browser instead
of calling out to the Python kernel?
dec0dedab0de wrote 1 day ago:
Since it was the ipython notebook I have only ever used jupyter
locally. So my first instinct is to do it all on the backend too, but
maybe they care about network latency? It looks like they're using
this as a collaborative tool. So kind of like a multiplayer online
video game, you don't want to waste time rendering graphics server
side.
niam wrote 1 day ago:
This looks great.
I do yearn for a day though when we're using something like Marimo over
Jupyter as a default for these kinds of things. Particularly in GIS
where there's more utility in being able to use a notebook-like
interface for an executable routine (rather than an analysis or
experiment, which is (and should probably remain) the primary use case
for Jupyter).
anentropic wrote 1 day ago:
There's something wrong with the scrolling on this page, like something
"puts the brakes on" (Chrome, macOS)
recursive wrote 1 day ago:
Drag with middle mouse button is supposed to scroll, but actually
does nothing on this page.
erk__ wrote 1 day ago:
It seems like something called "nicescroll" hijacks the scrolling on
the page, its compiled in the minified javascript so a bit hard to
tell though.
Turning off javascript seems to fix it though.
Stevvo wrote 1 day ago:
The Jupyter notebooks in ArcGIS pro are incredibly useful.
Unfortunately it's in Arcgis Pro. I'm thrilled to have the similar
setup that's not tied to a slow subsription software.
Coding assistants also work pretty well at doing GIS in python.
erremerre wrote 1 day ago:
I don't see any advantage working on JupyterGIS over working in QGIS.
cozzyd wrote 1 day ago:
If you're doing primarily analytics rather than making a map, I could
see how JupyterGIS might be a better choice.
(but you can always just use proj/GDAL at the import/export layer...)
fifilura wrote 1 day ago:
Notebooks is a nice and soft way to bring a programming mindset to
non-programmers.
Your actions are repeatable and can be stored in a centralized
repository.
There are probably some macro abilities in QGIS (it is an amazing
tool), but this means moving to "script first" from "click first".
twelvechairs wrote 1 day ago:
I can see it for programmers. Here you can use industry standard
python libraries (shapely, geopandas etc.). Nobody really wants to
learn PyQGIS (the python interface for qgis). So while qgis is much
more full featured for "desktop" gis (designed to compete with esri
arcgis) i can see the use case here for people who want to build
their own extensions and port code from this to other python projects
more easily.
Latitude7973 wrote 1 day ago:
I see a few advantages. For my work in particular, I have to rely on
creating desk study reports via exporting PDFs from QGIS - this
depends on export DPI, page size etc. Following that I have to pull
those plans into e.g. Word and it's a messy system.
A python notebook would be a nice way of generating reports of GIS
data in an interactive way without being forced to use pages, PDFs,
and embedded image files.
erremerre wrote 1 day ago:
Where I work, I can't give anything is not a word document to
anybody else in the company. A python notebook might help at
creating the figures for example, but I can already do that with
QGIS layouts.
If the working environment allows for checking/reviewing within the
notebook, I guess this could help automatise things.
tasuki wrote 1 day ago:
> Collaborative GIS Environment: Work together on geographic data
projects in real-time.
What does this mean? How is it collaborative in real-time? (I don't
even know how Jupyter is collaborative... as in, several people can
open a Jupyter Notebook and make changes simultaneously, and things
don't break for either of them?)
sph wrote 1 day ago:
Yes, Jupyterlab has a collaboration extension.
I've spent the past month writing a HTTP+WebSocket proxy to be used
with Jupyterlab for an ESA-adjacent company, and spent a good deal of
time trying to make the collaboration system work behind this proxy.
That said, I don't think my contribution is at all related to this
post.
lou1306 wrote 1 day ago:
I guess things might get nasty when you try to _run_ simultaneously,
unless they actually have implemented a CRDT for the whole notebook
state (it would still be nasty from a usability POV, but at least it
would be consistent). Just writing on the notebook would work just as
any other collaborative editor out there, I expect.
trungld wrote 1 day ago:
JupyterLab does use CRDT via yjs ( [1] ), and the JupyterGIS
collaboration system is based on the service provided by
JupyterLab. So multiple users can edit the file concurrently.
HTML [1]: https://github.com/yjs/yjs
arjxn-py wrote 1 day ago:
You can try JupyterGIS live on this deployment powered by JupyterLite -
HTML [1]: https://jupytergis.readthedocs.io/en/latest/lite/lab/
tasuki wrote 1 day ago:
Ok, you link to a deployment, and also the docs say this:
> Collaborative GIS Environment: Work together on geographic data
projects in real-time.
How does that work? I can apparently make changes to the files, and
even save them, but then my changes are gone when I reload it. Where
were the changes saved? This is exactly what I'm wary of when using
Jupyter...
skissane wrote 1 day ago:
> How does that work? I can apparently make changes to the files,
and even save them, but then my changes are gone when I reload it.
Where were the changes saved? This is exactly what I'm wary of when
using Jupyter...
The GP linked to JupyterLite, which is browser-based Jupyter based
on WebAssembly...
Then there is JupyterLab which runs on your own machine, or on a
server somewhere (e.g. a K8S pod)...
And then there are options like JupyterHub and Kubeflow which start
K8S pods for you dynamically to run JupyterLab in...
So "where is my data stored" all depends on how Jupyter is
deployed... at the moment, when I use Jupyter, I'm mainly using it
running inside a Docker container which in turn runs inside a K8S
pod (weird way of deploying it, but I have my reasons)... and then
the notebooks are stored in a volume (PVC) attached to the K8S pod
(statefulset actually), but that's just temporary storage while I
work on them, anything I want to keep permanently is put in Git and
pushed to our corporate Git host... and then our actual datasets
are mostly on S3 (or something else which speaks the S3 protocol)
tasuki wrote 1 day ago:
I don't know! They said "collaborative" and they provided a link.
I tried making some changes to test the collaboration and clicked
the "save" button. It changed the "last changed" to "now" and
upon reload my changes were gone.
It lied to me: my changes weren't saved! If I can't save a thing,
I want to be explicitly told so!
skissane wrote 16 hours 28 min ago:
Iâm pretty sure it saves it using browser storage APIs⦠so
rather than it lying to you, it could be your browser (or some
extension even) lying to it
TheChaplain wrote 1 day ago:
Question from the ignorant, how does this relate to OpenStreetMap?
pastage wrote 1 day ago:
Openstreetmap is not mentioned anywhere in the JupyterGIS repo [1].
As is mentioned OSM data can be used anywhere.
HTML [1]: https://github.com/search?q=repo%253Ageojupyter%252Fjupyterg...
arjxn-py wrote 1 day ago:
JGIS builds on OpenStreetMap and other open geospatial data sources
as basemaps or overlays. JupyterGIS itself is a Jupyter-based
interface for working with GIS dataâso you can visualize, query,
and process OSM or any other data directly from notebooks alongside
Python tools like GeoPandas.
gorbypark wrote 1 day ago:
This seems to be a Python based Jupiter notebook (style?) thing for
collaboratively working with GIS data/visualizations.
OpenStreetMap is a project to "map the world". In the end,
OpenStreetMap provides data (and map tiles) for other things to use.
Going out on a limb (since I haven't used it) but JupyterGIS can
probably make use of OSM data, along with other data sources.
harvey9 wrote 1 day ago:
Osm is a database (in the Wikipedia sense not the Postgres sense).
You will be able to use osm data in this tool.
rossant wrote 1 day ago:
Looks great! What visualization backend is used? Is it GPU accelerated?
arjxn-py wrote 1 day ago:
JupyterGIS uses OpenLayers as its visualization backend. It runs
entirely in the browser and supports GPU-accelerated rendering
through WebGL when available
DIR <- back to front page