had a thought around 5:11: I understand Linus' concept of not breaking backwards compatibility, so that `checkout` will continue to work even though there are newer commands that do the same thing, but if the idea is to have these new commands be "more semantic" than the old ones, why not at least show a message in terminal when using the old command to say "hey, there's a better way to do this now"...? surely that would help people who use Git every day (like me) to learn them more quickly.
I was trying to experiment with this feature and was confused why it wasn’t working without the interactive flag. Turns out I am still on 2.39. Thank you for the callout in the comments here so I know I’m not going crazy!
Hey Scott, I was barely able to get 50% of part 1 (some super heavy stuff for beginners) but this part was like you created this one for us only 🙌 the way you present these talks is just awesome 😂 please continue this series and let’s see a demo of gitbutler in the next part of this series and how’s about continuing this series until we don’t cover all 145 commands 😅😂
The most important thing to know about git : it is not a backup system, it is a collaboration system. You have to actually use backups with another software such as beyond compare, to reduce the risk of a bug in git that deletes files and folders. Sometimes you don't have 5 hours to become an expert at git and understand why it suddenly refused to save the modifications. The slightest mistake, misunderstanding, or misuse of a command can harshly punish you.
No need to become an expert - you only need to understand the fundamentals to be able to work confidently with Git. Watch Chacon's original introduction in order to really grok Git, and you'll find that it's quite difficult to get it to lose your data. Just set a branch before doing something tricky. If you also push that branch to some remote, it's virtually impossible to lose your stuff.
It’s version control, first and foremost. I have plenty of repos where I’m the only developer. I agree that it’s not a backup solution, i.e . if your disk is corrupted and you haven’t pushed your commits, you might lose data. However, as long as your file system is intact and you committed your change, basically anything “lost” can be recovered via git reflog.
Across the talks there have been several recommended git config -global options recommended. Is there a good place to go to see an exhaustive list of recommended configs based on new feature adds or even just a merge commit’s expert opinion on recommended config?
@@markusklyver6277 The man page for git config sucks for this (or at least it used to suck). It's huge and some of the more obscure options are documented extremely poorly (or even not documented at all).
Now onwards I will only use more domain specific/ explanatory commands instead of checkout I.e, switch and restore and allowing commit specific updates is also cool and yeah the best of all for jumbo repos I.e, scalar (any specific reason for not using git specific naming like git clone-heavy) 😂😎 I am feeling like if scalar might have introduced earlier then maybe pinterest never specify the refspec to reduce their clone time 🤪
GitButler looks really cool and very helpful, but unfortunately switching to a GUI client would change my workflow too much so I have to stick with either using worktrees or straight up separate directories for different branches.
Those YAML configurable hooks looks nice, can these be globalized? I didn't get it to work. With the attributes, that's a nice feature. I configured it so that in addition to calling exiftool, it also adds the hash (by making a script). This way you can see that it changed despite no change in exif data. You can also make it so git diff fires up the different pictures visually side by side, if you want to.
When working in JetBrains IDEs, I use Change lists: for example, I can create one for dev changes (that I definitely don't want to commit), other ones per branch. When commit, I select only needed change list(s) and other changes are not affected. Overall, JetBrains commit tab is quite comfortable for me.
I didn't see much in this talk that seemed of interest for me to look in to. I can see worktrees being useful. I already have multiple branches of a project checked out in to different directories but git doesn't know about the collection of directories. Using worktrees I can make it aware of the other branches being worked on.
The "restore" command semantically often makes less sense for me. I frequently check out changes I have never seen, stemming from other branches or even remotes, so it is definitely not restoring anything. So in many cases it fits, just "checkout" is more general.
That intro was almost as bad as Git. Use Mercurial or something else, Git is trash people think isn’t trash because so many people use it. It’s the JavaScript of change control.
Didn’t know Pedro Pascal is such a git nerd.
Amazing talk btw!!
😂
It's not every day you learn several new git features to try out. Thanks for a great talk.
thanks for watching!
The best kinda days
Yoo I was really excited by the first talk, so great to see the part 2 :)
Some reason I'm watching this part 2 :D !!
17:13 I am apparently one of the "three guys". I wished for this option for a long time. This is a huge lifesaver.
Agree, super useful.
Came here to say this! `--update-refs` where have you been all my life!
I love it! ❤
Presentation so fire you triggered the alarms haha. Loved both of these and I could listen/use many more :D
The course is undestandable and the intro song was so entertaining!
had a thought around 5:11: I understand Linus' concept of not breaking backwards compatibility, so that `checkout` will continue to work even though there are newer commands that do the same thing, but if the idea is to have these new commands be "more semantic" than the old ones, why not at least show a message in terminal when using the old command to say "hey, there's a better way to do this now"...? surely that would help people who use Git every day (like me) to learn them more quickly.
Damn, I would have appreciated knowing about worktrees about 2 weeks ago 🙃
Always a chance to learn something new
I've used fixup commits and rebasing for many years. Can't see how you'd get a clean, meaningful history without it.
Lovely talk as always!
The guy is about them beers hah
Nice hearing your talks, thanks for making them. Hopefully we'll have beers one day.
i love these talks
thanks!
I had no clue you don't need to do an interactive rebase to make autosquash work anymore... Nice to know :)
I believe this was _very_ recently added. 2.44 I think, so maybe a month ago?
I was trying to experiment with this feature and was confused why it wasn’t working without the interactive flag. Turns out I am still on 2.39. Thank you for the callout in the comments here so I know I’m not going crazy!
It's the best video I’ve watched the whole week. Thank you!
Woah, I didn't know about git worktree. That seems really useful and I've done it the poor man's way a bunch of times. Thanks!
These are awesome, thanks!
Aside from the large repository stuff (scalar) I would have put my hand up for each question, and that makes me happy 😊 Great talk Scott!
Dude these talks are so useful!!
Pretty glad I subscribed to the youtube channel :3
We're glad you did too! We'll try to make it worth your while. Let us know what you might like to see.
great idea to not repeat the talk 👍
Great presentation style! 👍🔀🚀
Hey Scott,
I was barely able to get 50% of part 1 (some super heavy stuff for beginners) but this part was like you created this one for us only 🙌 the way you present these talks is just awesome 😂 please continue this series and let’s see a demo of gitbutler in the next part of this series and how’s about continuing this series until we don’t cover all 145 commands 😅😂
great talk and great speaker!
`--update-refs` is great!
The most important thing to know about git : it is not a backup system, it is a collaboration system. You have to actually use backups with another software such as beyond compare, to reduce the risk of a bug in git that deletes files and folders.
Sometimes you don't have 5 hours to become an expert at git and understand why it suddenly refused to save the modifications. The slightest mistake, misunderstanding, or misuse of a command can harshly punish you.
No need to become an expert - you only need to understand the fundamentals to be able to work confidently with Git. Watch Chacon's original introduction in order to really grok Git, and you'll find that it's quite difficult to get it to lose your data. Just set a branch before doing something tricky. If you also push that branch to some remote, it's virtually impossible to lose your stuff.
It’s version control, first and foremost. I have plenty of repos where I’m the only developer.
I agree that it’s not a backup solution, i.e . if your disk is corrupted and you haven’t pushed your commits, you might lose data. However, as long as your file system is intact and you committed your change, basically anything “lost” can be recovered via git reflog.
Across the talks there have been several recommended git config -global options recommended. Is there a good place to go to see an exhaustive list of recommended configs based on new feature adds or even just a merge commit’s expert opinion on recommended config?
You can just read the git config man page.
@@markusklyver6277 recommendation (n): a suggestion or proposal as to the best course of action, especially one put forward by an authoritative body.
@@markusklyver6277 The man page for git config sucks for this (or at least it used to suck). It's huge and some of the more obscure options are documented extremely poorly (or even not documented at all).
worktrees are awesome
Now onwards I will only use more domain specific/ explanatory commands instead of checkout I.e, switch and restore and allowing commit specific updates is also cool and yeah the best of all for jumbo repos I.e, scalar (any specific reason for not using git specific naming like git clone-heavy) 😂😎 I am feeling like if scalar might have introduced earlier then maybe pinterest never specify the refspec to reduce their clone time 🤪
you didn't mention lefthook when talking about precommit and husky!!
nice intro)
GitButler looks really cool and very helpful, but unfortunately switching to a GUI client would change my workflow too much so I have to stick with either using worktrees or straight up separate directories for different branches.
Those YAML configurable hooks looks nice, can these be globalized? I didn't get it to work.
With the attributes, that's a nice feature. I configured it so that in addition to calling exiftool, it also adds the hash (by making a script). This way you can see that it changed despite no change in exif data. You can also make it so git diff fires up the different pictures visually side by side, if you want to.
If I had your public speaking skills I would be between 2x and 500x wealthier. Well done talk.
When working in JetBrains IDEs, I use Change lists: for example, I can create one for dev changes (that I definitely don't want to commit), other ones per branch. When commit, I select only needed change list(s) and other changes are not affected. Overall, JetBrains commit tab is quite comfortable for me.
I didn't see much in this talk that seemed of interest for me to look in to. I can see worktrees being useful. I already have multiple branches of a project checked out in to different directories but git doesn't know about the collection of directories. Using worktrees I can make it aware of the other branches being worked on.
Same here, I have like 5 different directories called git-1, git-2... I will try tomorrow
Why is Scott against a rebasing flow?
The "restore" command semantically often makes less sense for me. I frequently check out changes I have never seen, stemming from other branches or even remotes, so it is definitely not restoring anything. So in many cases it fits, just "checkout" is more general.
Deeeespaaacito
Man they really skirted the copyright with that 😂 cool how they turned it into a beat though.
Scotch Cone.
I was scared during the first few seconds of the video. Thought I was lost
it caught me off guard
Update refs❤❤❤❤❤❤❤❤😂
_ICT🌌1AM March 26th 2024_
Yeah, renase with references on top would be damn useful 2 years ago for me =)Now I just try not to use that approach
and yet, -q and --quiet still don't work consistently
almost nobody used autofixup? Wth, I'll do you one better, start using git-absorb
Nope. I don’t think that! 😂
Damn, coming at you with commit hooks 😢😢. Really bad stuff
hi 🎉?
hi
Nobody knows Git
That intro was almost as bad as Git. Use Mercurial or something else, Git is trash people think isn’t trash because so many people use it. It’s the JavaScript of change control.
OhNo, there's more 🫣