ponymontana🦄⚡ on Nostr: I invested a lot of time learning linux/bsd fine tuning my systems many times from ...
I invested a lot of time learning linux/bsd fine tuning my systems many times from ground-up. Now use daily my nerdy gentoo, compiling my kernel regularly with fine tuning for my machine, with only the feature that I use and extensive hardening and code-reduction. I compile my compilers regularly with security hardenings and stack protection, maintain my minimal fork of dwm for wayland and so on... In the end I builded the distro I wanted and didnt existed.
And I'm happy using it daily, the maintainance is all automated and really all works and dont need to loose time thinking too much everyday. The most of time/resources I loose are due to heavy long compilations, around 4 hours every mounth I think.
Things that require much more compile-time are GCC (~1:30h), glibc (~45mins), openssl (~45mins), linux kernel (~1h).
Pretty much this, maybe I'm missing something, but all other things compiles in max 30 mins, like python, go toolchain + rebuild all my go programs cause static libs, openssh...
I use binary for firefox and rust cause they are shit and dont want to compile. And because firefox would take at least ~3/4 hours.
I dont use GUIs, I dont need qt libs. I need just basic gtk gtk+ libs for basic things.
I use binary for monolithic heavy gui programs.
I also just put a binary when I need a jvm.
Thats it, I like it, I learned a lot diving into nerdy stuffs to make this thing works.
But today, if I will buy another machine and will need a new operating system, I'd absolutely go for a simple debian with some hardening and tweaks (like the kicksecure/whoonix approach).
Someone says "linux is free if you dont value your time" and this is half truth.
For the time and effort I put into this things I obtained in change a lot of experience and practice that I can use in other situations to makes machines work for me.
So I suggest anyone who want to learn more about these stuffs to go deeper, configure things manually, invest time in break things so then you can repair it. The value you can get from this approach is immense.
Then, at some point, real works need to be done, real things that makes life good require attention and life is too short to compile GCC forever just for the sake of.
So at some point you will know when is the moment to just go for things that works and have a "social convergence" of people. You will know when the "new time invested in deep nerding and auto-complicare your life" is not so valuable anymore in terms of "learning experience".
When this happens means its time to move on, keep what you learn with you and start nerding on other things, even other fields. Gardening this time? Nostr? Idk.
But after all I learned from my linux/bsd experience, today I'll go for debian as main os (or ubuntu/mint if my hardware need proprietary shit), where I dont need to think about nothing under the hood, knowing that is the standard.
And as all standard it is not perfect, nor the most secure, not the most fast and so on... but is reasonably good and a lot of people are thinking about how your system works for you, with you.
And I'm happy using it daily, the maintainance is all automated and really all works and dont need to loose time thinking too much everyday. The most of time/resources I loose are due to heavy long compilations, around 4 hours every mounth I think.
Things that require much more compile-time are GCC (~1:30h), glibc (~45mins), openssl (~45mins), linux kernel (~1h).
Pretty much this, maybe I'm missing something, but all other things compiles in max 30 mins, like python, go toolchain + rebuild all my go programs cause static libs, openssh...
I use binary for firefox and rust cause they are shit and dont want to compile. And because firefox would take at least ~3/4 hours.
I dont use GUIs, I dont need qt libs. I need just basic gtk gtk+ libs for basic things.
I use binary for monolithic heavy gui programs.
I also just put a binary when I need a jvm.
Thats it, I like it, I learned a lot diving into nerdy stuffs to make this thing works.
But today, if I will buy another machine and will need a new operating system, I'd absolutely go for a simple debian with some hardening and tweaks (like the kicksecure/whoonix approach).
Someone says "linux is free if you dont value your time" and this is half truth.
For the time and effort I put into this things I obtained in change a lot of experience and practice that I can use in other situations to makes machines work for me.
So I suggest anyone who want to learn more about these stuffs to go deeper, configure things manually, invest time in break things so then you can repair it. The value you can get from this approach is immense.
Then, at some point, real works need to be done, real things that makes life good require attention and life is too short to compile GCC forever just for the sake of.
So at some point you will know when is the moment to just go for things that works and have a "social convergence" of people. You will know when the "new time invested in deep nerding and auto-complicare your life" is not so valuable anymore in terms of "learning experience".
When this happens means its time to move on, keep what you learn with you and start nerding on other things, even other fields. Gardening this time? Nostr? Idk.
But after all I learned from my linux/bsd experience, today I'll go for debian as main os (or ubuntu/mint if my hardware need proprietary shit), where I dont need to think about nothing under the hood, knowing that is the standard.
And as all standard it is not perfect, nor the most secure, not the most fast and so on... but is reasonably good and a lot of people are thinking about how your system works for you, with you.