25:55 That's actually a pretty interesting idea. We use C/C++ as low-level standard for high-level programming languages and virtual machines. So why not using Git as a low-level framework for new version control systems, given its flexibility and power?
I think that's probably the point. Honestly, you want only the best and the brightest to be working on something as critical as the Linux kernel. That's the entire reason Linus Torvalds made it a point to not use C++ anywhere in the kernel: because it attracts too many shitty developers.
As with all software you need to learn how to use it and once you've done that you don't have to be an uber-genius to solve most of the day-to-day issues. The switching branches problem is solved by using --force The detached HEAD problem is solved by using 'checkout master' (or any other branch).
Well, if a hammer (or --force) solves the problem then why not use it? Using --force is actually pretty much the same as clicking "Yes" when you want to delete a file and your File Manager asks you "Are you sure you want to delete this file?".
Because it is inconvenient to use multiple commands for a behaviour that fits a very common workflow and that everybody would implicitly expect. Because of mixing concepts -- stash doesn't remember what got stashed on what branch, though it would be fairly simple to do. And finally, because one just forgets about stashed content while working on other branches. It would me much easier and safer if git just stashed changes automatically when switching branches and then automatically apply the stash when switching back.
25:55 That's actually a pretty interesting idea. We use C/C++ as low-level standard for high-level programming languages and virtual machines. So why not using Git as a low-level framework for new version control systems, given its flexibility and power?
6:34 The most annoying problem is switching versions under conflict
I feel sooo understood by this talk.
I wonder how Linux developers deal with these problems, since not all are as uber-genius as Linus Torvalds and Junio Hamano.
I think that's probably the point. Honestly, you want only the best and the brightest to be working on something as critical as the Linux kernel. That's the entire reason Linus Torvalds made it a point to not use C++ anywhere in the kernel: because it attracts too many shitty developers.
As with all software you need to learn how to use it and once you've done that you don't have to be an uber-genius to solve most of the day-to-day issues.
The switching branches problem is solved by using --force
The detached HEAD problem is solved by using 'checkout master' (or any other branch).
Well, if a hammer (or --force) solves the problem then why not use it?
Using --force is actually pretty much the same as clicking "Yes" when you want to delete a file and your File Manager asks you "Are you sure you want to delete this file?".
6:24 that dude needs to just stash the uncommited or unstaged work then switch branch easily. why whine about it?
Because it is inconvenient to use multiple commands for a behaviour that fits a very common workflow and that everybody would implicitly expect. Because of mixing concepts -- stash doesn't remember what got stashed on what branch, though it would be fairly simple to do. And finally, because one just forgets about stashed content while working on other branches.
It would me much easier and safer if git just stashed changes automatically when switching branches and then automatically apply the stash when switching back.
Git CLI is the worst on the planet!
Please give a few examples showing why you think it's unusable.
@@d5uncr Nobody said anything about git being unusable. It is just hard to use.
If you read Arun's and Piotr's comments they're more or less saying just that...
What a wonderful planet that must be (maybe behind its moon you could see some Mac "users" without any light shed upon them)
Git is simple you made it 100x hard.
Complicated with everything possible 🤡