I tried plotly a while back. Most viz tools are great when you have low/no customization requirements (e.g., streamlit was great too!). Unfortunately, custom stuff is still “easier” for tools that have a big historical community. At least, FWIW, that has been my experience…
No, I’m using Bokeh directly
Around once every 18-24 months, I find the need to create an interactive plot. I immediately think about adding interactivity to my Matplotlib plots using Panel/Hvplot because it’s been advertised as “easy" and, after about a day, I always come to the repeat conclusion that Bokeh is what I needed!
I didn’t realize that (food grade) white vinegar could be such an effective skin irritant. I was running some vinegar through my tankless water heater as a part of my annual maintenance tasks and I managed to spray vinegar everywhere thanks to a faulty hose. The cleanup was frustrating but it works!
Is Patchouli oil the new Axe Spray?! 🙅♂️ The scent (from others) is so strong that it causes me to hold my breath whenever somebody walks by with it 🤦♂️
You know, that cost multiplies exponentially when I’m reminded by my wife that, “Hey, you’ve been working on this thing for two hours now…⏳” 😅
When I add up the hourly rate along with the $300 in raspberry pi parts, that $600 M4 Mac Mini (beast of a base model) starts sounding more and more like an easy win that “Future ME” will thank me for 😅
😅 It’s a deeply satisfying feeling to combine a few CLI commands in order to “make it work” and the endorphins associated with being able to do seemingly impossible things. However, when it “stops working”, I always go back to check my “hourly rate” and quantify how much money I’m burning 🔥
tired: script kiddies
wired: chat kiddies
copy and pasting code you don't understand is not new, and not a recipe for longterm success
Having said that, I found myself questioning the value of the raspberry pi under higher loads as the total cost is converging around $300. This doesn’t include the cost of maintaining the hacky software/OS and the hours wasted for being my own IT admin 😅
Yes, I ran the diagnostics and everything passed. The read/write logs were consistent with the expected performance of the SD card. I’m running Homebridge with a lot of connected smart devices and so there is a lot of continuous logging that is slowing things down.
🥧 My relatively new raspberry pi5 was taking 10-15 minutes to fully reboot into the desktop GUI so I followed the advice to upgrade (from the microSD card) to an NVME SSD. 👍 Now it boots up in 40 seconds! The read/write speeds were tested before/after and it is around 10x faster for both 🚀
Buying glasses/lenses from Costco is really, really hard to beat and basically makes vision insurance obsolete. I even bring my own frames and it is still cheaper than paying into insurance.
I continue to be amazed by how powerful, versatile, & complementary the Bash+Python combination is! It works across multiple platforms & can accomplish most automation tasks with very little code/effort. They can stand alone or you can call a Python script from within a shell script, & vice versa 😍
One question I have is the open source (Python package) contributor use case. If I have a wide range of contributors who may use `conda` or `venv` (or now Pixi) to setup their development environment, is there a good way to support that with Pixi? Or am I forced to only support Pixi contributors?
I swear the Brother motto should be, “Brother printers. They’re fine”
www.theverge.com/23642073/bes...
I actually drove to Office Depot to look at them in person. The first 10 minutes was me trying to figure out why I was starting at $250+ printers with scanners, only to realize that they were for biz. The home printers were two aisles down, almost to trick suckers into buying the expensive ones.
Thank you. I wish there was a way to configure it so that it will block if the pi-hole service is “up” but fail over to the secondary DNS if the raspberry pi’s address cannot be resolved/reached
I had to look behind me after reading this message to make sure you weren’t standing right behind me (yes, it was a Brother printer!!) 😅
I am slowly familiarizing myself with Pixi by the @prefix.dev team
Since I use my raspberry pi (and pi-hole) as my DNS, every time I reboot the pi my entire network goes down until it is fully booted back up (sometimes 10-15 minutes 😅). Today, I realized that I could set a secondary DNS (the pi is my primary DNS) and there is little to no interruption now!
Since I rarely need to print things at home, I had been using a 13 y.o. laser printer and connected to a wired print server for several years. It recently crapped out and I was really surprised to find a new wireless (airprint) replacement for $100. Got it up and running in less than 20 minutes!
Hey, turns out you CAN use CUDA packages from @conda-forge.org for compilation, the docs just needed to be updated. Now they are! conda-forge.org/docs/user/fa... Big thanks to everyone who helped me get all the CUDA info correct in github.com/conda-forge/... and special thanks to John Kirkham. :)
@kim-chen.bsky.social 👋
Struggling with extracting text from PDFs? I explored Nougat-OCR & Docling to improve text extraction. Discover how they handle equations, tables, and figures. Share your thoughts! Retweet if helpful. https://buff.ly/4iZoHtc #pdfextraction #researchtools
What are your favorite data science conferences to attend? I’m looking for suggestions for 2025! #dataBS
NUCS is a Numba-powered constraint solver: github.com/yangeorget/n...
It's an area I don't know much about, but Optimal Golomb Ruler stuff sounds cool!
(A long time ago I think I participated in the distributed.net search for OGRs)
🙋♂️ Does anybody have any tips for searching for users on here? Even when I follow them, I am unable to @mention them (using their full handle) and, instead am seeing “No results” 😢
The #SciPy2025 Call for Proposals is live (as of mid December 2024, sorry forgot to share) over on scipy2025.scipy.org! Deadline for submissions is 2025-02-26, so get submitting!