Benoît Bovy

🚨 Excited to announce a new release of ipytone, a big one and the first “public” release! More details below 🧵 But let’s start with this little piece of music made entirely in a @ProjectJupyter notebook (+ some “waving topography” made with pyvista / ipygany). 🔊🔊🔊🌍 1/N

I can't say I've been very active on social networks so far... Here I won't be at all. You can find me there: bsky.app/profile/benbov… @benbovy" target="_blank" rel="nofollow noopener">fosstodon.org/@benbovy

🚀 We are thrilled to announce the release of the Icechunk storage engine, a new open-source library and specification for the storage of multidimensional array (a.k.a. tensor) data in cloud object storage. Read our blog post about Icechunk here: earthmover.io/blog/icechunk

@choldgraf @rabernat Yes in this case “confusingification” would be a better term.

@rabernat agree - I think we should definitely be paying companies for this stuff and I generally don't like the investor-subsidized pricing model anyway because it has long-term strategic downside (see also: enshittification). IMO, biggest problem here is clear communication.

During my time at @Columbia, I'm proud that we partnered on grants with @anacondainc folks like @pwang, @JamesABednar, and @mrocklin. I would have been even happier if our IT department had paid a large annual fee to support their work! 7/

@MilesCranmer @fcrameri ‘s color maps fabiocrameri.ch/colourmaps/, matplotlib cmaps pypi.org/project/cmcram…

Matplotlib palettes are useful for science, because they show contrast well, but I find them not to be visually pleasing. Are there any good libraries with aesthetics-focused matplotlib colorschemes?

@choldgraf @anacondainc The problem is about enforcement: it is still so easy to download packages from the anaconda channel and not be aware (or forget about) that change of license… I remember the situation was quite messy after the org was contacted by Anaconda about that.

@choldgraf @anacondainc It happened a few years ago to the research institute I work with. TBF, I think it only applies to downloading packages from the anaconda channel, no license issue with conda / conda-forge.

Is @anacondainc threatening to sue non-profits and universities? Does anybody know more about this? theregister.com/2024/08/08/ana…

@trevmanz @TEGNicholasCode @xarray_dev Could be a nice (and pretty easy?) addition to github.com/benbovy/xarray…, which already uses #anywidget.

@TEGNicholasCode @xarray_dev let’s chat :)

Watching this anywidget talk by @trevmanz - incredible work! This "quak" widget for interactive exploration of a database via duckdb... We NEED that but using @xarray_dev calls instead of SQL

@wuoulf @condaproject Downloading repodata has been the main bottleneck since fast spec solving in my experience, especially when I just need to install one or two packages in an environment. Speeding up this step will be a great improvement!

We're working on a feature to make the @condaproject experience a lot faster for big repositories - "sharded" repodata (so that initial downloads become a lot smaller). Hoping to ship an experiment soon!

@DPManchee It is a small Python module included in the fastscape-demo repository (github.com/fastscape-lem/…). It requires xarray-spatial. The conda env file w/ all dependencies for fastscape-demo is here: github.com/fastscape-lem/…

@DPManchee Have you tried mamba? github.com/mamba-org/mamba

@DPManchee If you don't really need to build the library from source and use it via the Python bindings, I strongly recommend installing it using conda, e.g., "conda install fastscapelib-f2py -c conda-forge". Fastscape (the python package) uses it under the hood (also available w/ conda).

@PhilippJFR @kolibril13 @davidbrochart @Panel_org @marlene_zw I also find "used by" on github a pretty nice and complementary metric indicating some real world usage, although it doesn't seem to always show up properly.

@kolibril13 @davidbrochart @Panel_org @marlene_zw Indeed, downloads definitely have their own problems as a metric. There really isn't a great proxy for real world usage, but downloads at least are harder to game and indicate something real.

Thanks for your support. Means the world.

@JMTurowski Maybe @zzsylvester's segmenteverygrain? github.com/zsylvester/seg…

Geoscience twitter: I want to automatically measure grain size distributions from thin section photographs. Any recommendations of software?

@TEGNicholasCode Great talk!

Slides posted here! speakerdeck.com/tomnicholas/cu…

I'm talking about Cubed tomorrow - come and hear about this clever framework for processing large array workloads using serverless functions + @zarr_dev !

@DrFionaClubb @GIStuart I wish I could attend the workshop (and also share a bit about the hard work we've been doing on @fastscape over the past few years), but unfortunately I had a schedule conflict those two days.

I am hiring! I am looking for a PhD student to work with myself, @ChrisAndermann, Micha Dietze and Niels Hovius to study the connectivity between groundwater, surface water, and sediment yield. The job will involve code development and field work in the beautiful Eifel mountains.

@davidbrochart My 2-cents: I like the idea but I’d remove the antialiasing effect completely or limit the number of colors (e.g., to match 256 colors terminals). Also the font looks a bit weird and out of scope, is it a monospace font?

Before/after, what do you think?

@ProfMoucha In theory you can use it in both cases. For multiple valleys it is possible to set any inner node of the grid as base level. Fastscapelib doesn't provide anything like a function to compute Hack's law, though, but it is straightforward to implement.

It took a long time, but finally I’m glad to announce a new release (re-written from scratch) of Fastscapelib: a core library for landscape evolution modelling Some highlights 🧵 (1/8)

@opengeospatial @rabernat @pygeoapi Does the OGC API - Coverages supports selecting multiple points at once? In the draft specs I see that "subset" supports either one point or an interval, but I might be missing something?

@rabernat @pygeoapi Hi Ryan. The OGC API - Environmental Data Retrieval standard (ogcapi.ogc.org/edr/) or OGC API - Coverages candidate standard (ogcapi.ogc.org/coverages/) would both work for this use case.

Question for the @opengeospatial / @pygeoapi #geospatial crowd. Say I have a raster datacube (lat, lon, time) and I want to provide an API to sample discrete lat / lon / time points from this space. What is the right OGC API to use here?

Thanks to all contributors to this release!! @HSophiaDavies, Guillaume Cordonnier, @AdrienDelsalle and people at @QuantStack for their contribution and advice on the design of the library (also for #xtensor!) It is supported by @GFZ_Potsdam (Jean Braun’s section @GFZ_SurfMod).