8 minutes read | 1549 words by Ruben BerenguelSome links are affiliate links
Spark 3 is here! Rejoice!
NOTE: Spark (quite a bit), Python, Golang, Maths and a bit of miscellanea. Expect a similar wide range in the future as well. You can check all my weekly readings by checking the [tag here](https://www.mostlymaths.net/search/label/ReadingsOfTheWeek, Readings). You can also get these as a weekly newsletter by subscribing here.
This is what we drink at home during summer (I’m sipping a glass right now). This makes it sound more complicated than it should. Steps:
Buy dried hibiscus flowers and wait until they are delivered
Place enough to cover the bottom of a glass bottle or heat/cold proof container (say, a 1 or 2 liter bottle) so you don’t see much of the bottom (tweak as needed after the first time you do)
Boil 1/2 cups of water and pour them on step 2
Let it rest until warm, fill fully and place in the fridge
Enjoy, optionally with sugar.
If the flowers are large (which is the usual) you just need a bottle with a coarse filter, or just be a bit careful. No need to “do” much. It’s excellent for the heat, and if you happen to have pomegranates around, get a thin sliver of peel (without no white rind, important) and add to step 2. Optionally as well, add a lot of peel and some of the pomelo flesh for a bitter, tarter taste.
I use Databricks at work (and know a few people there). It is an excellent product, but… are open-source-turned to companies really going to be great/shape the future/not close? There are not that many examples to draw from, and most I can think are either very recent (Confluent, Databricks) or not very large/doing well (Lightbend). And that’s not counting Hortonworks/Cloudera.
This is a very understandable explanation about how to use automated differentiation (a very powerful technique) to optimise. And in particular, how to use the JAX library (an almost numpy drop-in replacement with built in automated differentiation in CPUs and GPUs).
The harmonica is one of the instruments I own and play badly, and this is a very interesting tour through its history and renaissance. I also recommend you get one (they are very cheap) and learn to play some blues. Why not? Music is fun.
We have experimented with Dask at work recently. We have liked what we have seen… But in the end we are defaulting to Spark/Pyspark/Pandas or plain ol’ from multiprocessing import Pool. Still, dask is good to have in your bag of tricks.
Times are a-changing, but in what I know from Europe, even college kids at that time could cook. We still have grills, though, but we call them sandwich-makers, and use them for that. Usually with Nutella.
This is the first chapter in a 7-post long tour across modern practices with Python. This one is about setting up your device, use pyenv to install different versions of Python, using Poetry to manage dependencies and create a simple CLI application usinc Click and requests to fetch from an API.
Every time I think I don’t really need to learn Rust I’m reminded of Weld. Although Weld is only going to be an internal piece in other frameworks, I need to understand how the sausage is made. You can read the slides here
You can find a transcript here. This is an excellent introduction to property based testing and the Hypothesis Python library. And about advanced testing concepts (like metamorphic testing). Very high on my recommendation list this week.
I’m not a fan of Robert Lang‘s origami design (too complex, too many insects), but he’s not only a master at it but is also very good showing why. He has several books explaining how to construct very complex bases and crease patterns, and even wrote software for that. Very recommended video.
Another book by the Heath brothers, and like others I have reviewed before, recommended. They are always entertaining and to the point, even if you have read/heard about the matter several times and know all the stories.