Hey souenzzo - it should repeat the same exercise when you fail in Practice Model. If you want it to mimic Simon, just set "Show Last Note" in the Settings.
I mean.. how is this different from any OS distribution? Apple can push whatever. So can Red Hat or Ubuntu or Gentoo. Unless im literally running Linux From Scratch im at the mercy of maintainers to do whatever they want.
I'm not sure what the current state of most distributions is, but I remember update applications providing an option to accept or reject individual packages. Even without that, you could preview the list of pending updates and delay them indefinitely, do manual updates of individual packages, or configure it to ignore particular packages during updates. Historically, I believe that you could block certain updates on Windows as well - or maybe you could just rollback and update. Of course none of this is considered user friendly so things may have changed.
But where does the original compiler come from? Reproducible builds are only as good as the compiler used to compile them. That's the point of Trusting Trust. If you build with a backdoored compiler and I reproduce your build with the same backdoored compiler, that solves nothing. This is why full-source bootstrap is important[0].
It would be very very hard to actually accomplish something like that on mainstream x86/arm compilers. And hide it from every debugger in the world. If it diminishes the value of reproducible builds, it's by something like 1%.
> Reproducible builds are only as good as the compiler used to compile them.
Which is so so so much better than "as good as nothing".
There are a lot more distros than RH, Ubuntu, Gentoo and LFS. And none of them will show you ads except maybe Ubuntu. Plus you can also look at *BSD.
None of them comes close to what Microsoft is doing. To me, your comment looks like you do not understand the Linux eco-system. Plus IIRC, LFS can now come with compiled binaries.
> Apple can push whatever. So can Red Hat or Ubuntu or Gentoo
In the case of Ubuntu and Debian, and to a lesser extent RedHat, I trust the developers not to do that because they have a history of not "just pushing whatever".
Also in many cases I actually know these developers, and I can go round and ask them / remonstrate with them / put a brick through their window / other response if required about it.
"Ubuntu will apply security updates automatically, without user interaction. This is done via the unattended-upgrades package, which is installed by default."
Right, but it's a minor annoyance, get rid of it with:
sudo apt-get remove --purge unattended-upgrades
(doesn't trigger removal of anything else, and you'll enjoy 420kb of additional disk space).
OTOH the real issue with Ubuntu is snap(d). Snap packages definitely do auto-update. You may want to uninstall the whole snap system - it's (still?) perfectly possible, if a little bit convoluted, due to some infamous snaps like firefox, thunderbird, chromium, or eg. certbot on servers
Or just use Debian or any snap-free fork for the matter.
I remember hearing about people doing that around the time that windows 98 was still current. It was really impressive.
At the time, the idea of an operating system using a gigabyte of space was a fantasy to most people. Now, I wonder when Microsoft Windows will pass the terabyte threshold.
I find that hard to believe, I also don't think LLMs provide the value some others are seeing but there is also code search and refactoring tasks that LLMs can help with. Instead of heaving to write a codemod, I can just write a prompt and accomplish the same thing in much less time. To say they have literally zero value just seems uninformed tbh. That's the sort of black and white thinking that isn't very helpful in the conversation.
Postgres or Supabase MCP is really useful for me, especially when dealing with data related issues like permissions bugs. It seems faster to me. Example:
> I cannot see the Create Project button as user project-admin@domain.com. Please use Supabase MCP to see if I have the correct permissions, if so, are we handling it correctly in the UI?
reply