| web | https://sgillies.net |
| github | https://github.com/sgillies |
| web | https://sgillies.net |
| github | https://github.com/sgillies |
Post-Austin/CAST I'm touring #ASynesthetesAtlas through AZ, NM, & CO. First up: Sat, 4 Apr 2026 in #Tucson @ #WaveArchive, a loaded program w/locals #JeffreyGordonEvans, #HannahJoyce , & Steev Hise (@detritus ) in various configs.
#experimental #improvisation #expandedCinema
#maps #cartography #OpenStreetMap #foss4g
Need to pick a sample of size K from a stream of events of unknown size? You need reservoir sampling. The article below introduces the basics, and a new (maybe?) variation that's useful when evenly spreading out samples is helpful
----
Alternative intro: The commercial profiler I built is no longer available; might revive it someday, who knows.
But I did learn some generally useful things from building it. Here's one—a variation on reservoir sampling that smooths out the gaps in profiling. This is helpful for timeline-based visualizations where you want samples to each represent the same amount of time.
----
https://pythonspeed.com/articles/reservoir-sampling-profilers/
Well hello there 🌱
In my frugal era, using toilet roll tubes as starters for my seedlings.
I wonder how many companies put all of the risks of the closed source coding agent collecting loads of data about corporate laptops on their risk register or accounted for it in their compliance processes.
I’m guessing they had the chatbot handle the details.
https://www.theregister.com/2026/04/01/claude_code_source_leak_privacy_nightmare/
Tech journalism has been dead for a while. I hope we can reverse this trend. If not, individuals will have to continue to step up: