Thanks for sharing, well done on reverse engineering the prices API.
I live in Poland and did a similar project tracking fuel prices, however we don't have any data sources with reliable station pump price data, so I ended up tracking the trend and sending signals whether to wait or refuel now.
I'm estimating prices based on brent barrel, usd rate, crack spread, wholesale prices (luckily we have local Orlen monopoly), some google maps crowd source data, scraping numbers from fuel station livestreams etc. I built a numeric model that gives 80-92% accuracy on price predictions in the horizon of 1-7 days.
That was fun a project and ppl find it useful.
Interestingly, my data analysis showed how significantly government decreased crack spread of Orlen before last parliament elections..
It reminds me of a project I made in 2017 to control a drone using voice commands. Alexa skills paired with IoT messages made it react in less than 100ms. Cool stuff, but useful as a chocolate teapot.
We have done automated utility inspections for the past few years. Computer vision is not a problem any more. When we were starting we had to annotate thousands of images. Today some use cases require less than a 100. More interesting problems are pre-field planning, sagging, or BLOS compliance
Yes, and hundreds of developers just spent 5 minutes reading about it.
This investment makes sense for OP who opens hundreds of terminal sessions, but does it make sense to bother about 1s delay when they start the terminal session once in a while? I have several sessions opened all the time, the last time I wasted my 1s on zsh was in 2025 when I rebooted my machine due to a nvidia drivers update.
Interestingly, my data analysis showed how significantly government decreased crack spread of Orlen before last parliament elections..