Generate endless markov nonsense for LLM crawlers to choke on. Basically the young kids (LLMs) being forced to listen to grandpa Markov's senile babbling over an excruciatingly slow dinner..
Two gripes about on the cube rule; It doesn't readily differentiate between topology of a dish and a single serving. It could very well add more dimensions to the identification model besides topology, there are plenty of other factors that define the portability and experience of eating a food (let's face it, that's what the debate of identification was really about this whole time.)
I would highly recommend the History of English Podcast. This particular observation made by OP is thoroughly covered in this particular episode: https://youtu.be/T0ED-FV7O50
Even setting up a vlan doesn't work half the time because the mobile apps don't talk directly with the appliance but phone home to a cloud service. A cloud service that will eventually go offline and leave the appliances orphaned. That's how GE's thermostats work.
Someone should make a service like this except you actually upload the file directly to the other user rather than uploading it to a 3rd party (encrypted or not). Yes I get you would have to wait for the user on the other end to connect to you before the transfer starts but if you're uploading 100+GB of data you're going to be leaving that browser tab open for awhile anyways.
I think you're underestimating how impossible a task it is for China to hit a moving target. Even so, their move towards isolationism is at odds with an industry that has the most complex and globally integrated supply chain in existence.
I do use helix but haven't taken advantage of the git integration. Maybe I'm unaware of its power.
For fish, I defined my own fish_prompt function with an indicator if there are uncommitted changes. It's just running git status under the hood. I have a TODO in that function to run a pijul diff in the directory if git status returns nothing...
The 1.0 is in beta. There has been a lot of refactoring to get it to this point. I would say there's still many quality-of-life features missing that would stop me from using it in a professional setting but for hobby projects it's meeting my needs (and gets better with each new beta build). They only have a few project backers but the main developer has been working very steadily on it.
This is actually why I prefer using pijul. I don't want to commit my secrets to a git repo and nix will refuse to build because I'm pulling in files that aren't tracked. Simple solution is to not make the flake directory a git repo and it won't complain. That's my solution at least. I also prefer using git (and therefore pijul) via cli rather than as a text editor integration so my experience differs.
I use it for self hosting because all I need installed is sshd and the pijul package. Then I can set my server's
;p as my remote. The "nest" web UI (the Pijul equvivalent to git tea) is in development and not open source yet, but you can use the hosted version at https://nest.pijul.com/ if you're curious.
Is that a yes or no question, or merely a suggestion?