pletnes

17 ms·

271.
by pletnes 4y ago
Python keeps growing in number of users because it’s easy to get started, has libraries to load basically any data, and to perform any task. It’s frequently the second best language but it’s the second best language for anything. By the tim
272.
by pletnes 4y ago
Btrfs does stuff like instantaneous snapshots for a multi terabyte size filesystem. I have no idea how they do it, but apparently it’s 100% possible.
273.
by pletnes 4y ago
Yes, and with both supporting arrow, you can zero-copy share data from polars into pandas once you’ve completed your initial data load+filtering.
274.
by pletnes 4y ago
I have some thinkpads privately and an M1 MBP at work. Bang-per-buck is best with the MBP, no doubt. The absolute number of bucks for a used thinkpad is super low and they live for a long time, so I have old thinkpads for light home use. Th
275.
by pletnes 4y ago
FastAPI! It shows the thinking and development process with diffs, going through each task. (There are other projects with better «reference style» docs)
276.
by pletnes 4y ago
There’s a chart of the speed - it’s around 250 knots so they’re going fast enough (source: personal MS flight simulator experience). The angle of attack is not shown, I think.
277.
by pletnes 4y ago
I don’t agree with the title, but this article is great if you want to get to the very bottom of the floating point rabbit hole. https://people.cs.pitt.edu/~cho/cs1541/current/handouts/gold... What Eve
278.
by pletnes 4y ago
I’ve done C++ in some projects and still think 3 months re-training might be necessary for new language features, and old foot-machine-guns for that matter! C++ is a dangerous beast if I ever saw one.
279.
by pletnes 4y ago
My moonlander took a few days/weeks of getting used to, but it’s not very «far out» in my mind. The biggest hurdle is getting used to the ortholinear keys, but this quickly feels much better. Would warmly recommend to anyone spending m
280.
by pletnes 4y ago
Another point is the massive neutron flux, which will get some nuclear reaction out of the bomb material in and of itself. Neutrons will set off both fission and fusion reactions, simultaneously.
281.
by pletnes 4y ago
What is then, in your opinion, an actually good ETL stack? Many use Spark/Databricks/Synapse, which has all 3, but these seem bulky and needlessly complicated for many smaller projects. Even TB sized datasets don’t need Spark, in
282.
by pletnes 4y ago
Surely, AutoHotKey should be the default. Most «data moving people» are not programmers. Okay, maybe that’s «data pipelines», not «data engineering pipelines», but…
283.
by pletnes 4y ago
No flak intended, AHK is much easier for non-programmers and can do more than you’d think. I’ve show a few people how to rig up an automation in a few minutes. The bar to entry is much lower than for any of the «serious» languages out there
284.
by pletnes 4y ago
Python buys into pypi, which has a 6-digit number of packages, e.g opencv can read your screen to find buttons, etc.
285.
by pletnes 4y ago
Pyautogui? https://github.com/asweigart/pyautogui That one is also cross-platform, if that’s useful to you.
286.
by pletnes 4y ago
I just downloaded v2, purely by incident (haven’t used AHK for some time) but there is now official support for compile-to-exe. No idea how good it is, I’ve only used AHK for one-offs, short term projects, etc.
287.
by pletnes 4y ago
I believe duckdb does not load the whole csv file into memory. It will load a few rows to find column headers and guess data types.
288.
by pletnes 4y ago
Cool! Although I believe duckdb can do it on disk / out of memory, so querying huge files are possible. I also like its syntax, I tend to CREATE VIEW mycsv AS SELECT * FROM ‘my.csv’ (or similar). Then I think you can select or join ev
289.
by pletnes 4y ago
Have a go with duckdb next time - you can query csv files without loading them first.
290.
by pletnes 4y ago
It’s got plugins for some stuff Excel don’t. I’ve used one for map/gis data to look at lat/long data on a map. Never seen that in Excel. The dataset was too big for Excel, too.
291.
by pletnes 4y ago
My impression is that many power users (not web developers) want to download datasets. Sometimes they’ll use a browser to grab a csv or json file, then stuff it into excel/R/python/… to do some sort of analysis work. This kin
292.
by pletnes 4y ago
Some vim plugins are slow. Fortran highlighting used to be super slow for certain use cases 10 years ago. I think there were some bad regexes. Latex documents are often long (cf other code) and the syntax rules are complex.
293.
by pletnes 4y ago
Anaconda python and activestate python are products in this space, I’d argue.
294.
by pletnes 4y ago
Where is that released? Would be good to know.
295.
by pletnes 4y ago
As can bash and zsh!
296.
by pletnes 4y ago
I don’t know where to start when the script is in ./Scripts/ or ./tools or ./bin or ./shared/tools. Make has a convention that its file is called Makefile. Just has Justfile. Easy to find. Justfile - here I don
297.
by pletnes 4y ago
Make is a mess to install on windows. Just just works.
298.
by pletnes 4y ago
Don’t forget discoverability - writing «make» and whacking tab gives you a nice menu of what can be done with the Makefile. (Just, too.)
299.
by pletnes 4y ago
Having tried all 3, I’ll warmly recommend WSL2 instead, which is basically linux. But many people do work on windows, even write software for it. While I’m in your Linux based corner, some people aren’t, and it’s nice to share a task runner
300.
by pletnes 4y ago
Just runs fine on windows, too, which most people seem to miss. Bash is not always available, make even less so.

More ›