When you stare into git, git stares back.

It's fucking infinite.

Me 2 years ago:
"uh was it git fetch or git pull?"

Me 1 year ago:
"Look, I printed these 5 git commands on a laptop sticker, this is all I need for my workflow! branch, pull, commit, merge, push! Git is easy!"

Me now:
"Hold my beer, I'll just do git format-patch -k --stdout HEAD..feature -- script.js | git am -3 -k to steal that file from your branch, then git rebase master && git rebase -i HEAD~$(git rev-list --count master..HEAD) to clean up the commit messages, and a git branch --merged | grep -v "\*" | xargs -n 1 git branch -d to clean up the branches, oh lets see how many words you've added with git diff --word-diff=porcelain | grep -e '^+[^+]' | wc -w, hmm maybe I should alias some of this stuff..."

Do you have any git tricks/favorites which you use so often that you've aliased them?

I once reviewed a Pull Request made by a fairly junior developer. They had joined recently, and this was one of the first times they had to touch a bigger part of the code.

Due to a mix of inexperience, new (to them) coding standards and lack of git knowledge, they ended up with a mess of a PR, with a few thousand lines changed, and no way to split it off.
I ended up spending the best part of a day reviewing the whole thing and requesting changes.

Even with the long list of improvements, however, I wasn't sure they would get the magnitude of their fuckup.
So I decided to use a real-world, palpable way to show them what they had done: I went and printed the github diff for that PR. It rendered the glorious amount of 73 pages.

I'll never forget their face, and those of their teammates, when I barged into the room with a thick wad of paper and deposited them on their desk.

At least it worked. I never saw another big, ill-thought pull request from them again.

I love #VisualStudio.
See how visual studio really makes easy to analyze difference between two version of file. 😘😘

So I'm back from vacation! It's my first day back, and I'm feeling refreshed and chipper, and motivated to get a bunch of things done quickly so I can slack off a bit later. It's a great plan.

First up: I need to finish up tiny thing from my previous ticket -- I had overlooked it in the description before. (I couldn't test this feature [push notifications] locally so I left it to QA to test while I was gone.)

It amounted to changing how we pull a due date out of the DB; some merchants use X, a couple use Y. Instead of hardcoding them, it would use a setting that admins can update on the fly.

Several methods deep, the current due date gets pulled indirectly from another class, so it's non-trivial to update; I start working through it.

But wait, if we're displaying a due date that differs from the date we're actually using internally, that's legit bad. So I investigate if I need to update the internals, too.

After awhile, I start to make lunch. I ask my boss if it's display-only (best case) and... no response. More investigating.

I start to make a late lunch. A wild sickness appears! Rush to bathroom; lose two turns.

I come back and get distracted by more investigating. I start to make an early dinner... and end up making dinner for my monster instead.

Boss responds, tells me it's just for display (yay!) and that we should use <macro resource feature> instead.

I talk to Mr. Product about which macros I should add; he doesn't respond.

I go back to making lunch-turn-dinner for myself; monster comes back and he's still hungry (as he never asks for more), so I make him dinner.

I check Slack again; Mr. Product still hasn't responded. I go back to making dinner.

Most of the way through cooking, I get a notification! Product says he's talking it through with my boss, who will update me on it. Okay fine. I finish making dinner and go eat.

No response from boss; I start looking through my next ticket.

No response from boss. I ping him and ask for an update, and he says "What are you talking about?" Apparently product never talked to bossmang =/ I ask him about the resources, and he says there's no need to create any more as the one I need already exists! Yay!

So my feature went from a large, complex refactor all the way down to a -1+2 diff. That's freaking amazing, and it only took the entire day!

I run the related specs, which take forever, then commit and push.

Push rejected; pull first! Fair, I have been gone for two weeks. I pull, and git complains about my .gitignore and some local changes. fine, whatever. Except I forgot I had my .gitignore ignored (skipped worktree). Finally figure that out, clean up my tree, and merge.

Time to run the specs again! Gems are out of date. Okay, I go run `bundle install` and ... Ruby is no longer installed? Turns out one of the changes was an upgrade to Ruby 2.5.8.

Alright, I run `rvm use ruby-2.5.8` and.... rvm: command not found. What. I inspect the errors from before and... ah! Someone's brain fell out and they installed rbenv instead of the expected rvm on my mac. Fine, time to figure it out. `rbenv which ruby`; error. `rbenv install --list`; skyscraper-long list that contains bloody everything EXCEPT 2.5.8! Literally 2.5 through 2.5.7 and then 2.6.0-dev. asjdfklasdjf

Then I remember before I left people on Slack made a big deal about upgrading Ruby, so I go looking. Dummy me forgot about the search feature for a painful ten minutes. :( Search found the upgrade instructions right away, ofc. I follow them, and... each step takes freaking forever. Meanwhile my children are having a yelling duet in the immediate background, punctuated with screams and banging toys on furniture.

Eventually (seriously like twenty-five minutes later) I make it through the list. I cd into my project directory and... I get an error message and I'm not in the project directory? what. Oh, it's a zsh thing. k, I work around that, and try to run my specs. Fail.

I need to update my gems; k. `bundle install` and... twenty minutes later... all done.

I go to run my specs and... RubyMine reports I'm using 2.5.4 instead of 2.5.8? That can't be right. `ruby --version` reports 2.5.8; `rbenv version` reports 2.5.8? Fuck it, I've fought with this long enough. Restarting fixes everything, right? So I restart. when my mac comes back to life, I try again; same issue. After fighting for another ten minutes, I find a version toggle in RubyMine's settings, and update it to 2.5.8. It indexes for five minutes. ugh.

Also! After the restart, this company-installed surveillance "security" runs and lags my computer to hell. Highest spec MacBook Pro and it takes 2-5 seconds just to switch between desktops!

I run specs again. Hey look! Missing dependency: no execjs. I can't run the specs.

Fuck. This. I'll just push and let the CI run specs for me.

I just don't care anymore. It's now 8pm and I've spent the past 11 hours on a -1+2 diff!

What a great first day back! Everything is just the way I left it.

a tale of daily frustration:

git fetch
*yup I'm up-to-date ...*
git add -p .
*hack in beautiful patch ...*
git status -bs
*correct branch, didn't forget any files ...*
git diff --cached
*yep, that is what I mean to commit ...*
git commit -m"[TKT-NUM] Meaningful commit message"
git log -p -1
*double-checking ... looks good ...*
git push remote tkt-num-etc

*for a brief moment feel accomplished ...*

*notice typo in commit message ...*

I don't have a funny image or punchline to sum this post up. But know that if you recognise this feeling, then I am your brother in git.

Can anyone make a simplistic version control like git for microsoft word? I just want an easy check in, check out, and seeing the diff. Goddamn some of these tools got a lot of steps

Boss: some consultants worked on this feature extending some legacy code
Boss: it's 90% done
Boss: they used FTP. It uses iframes and we fired them when they couldn't get the frontend modules working in sync with the backend.
Me: git checkout -b herewegoagain
git diff-tree --no-commit-id --name-only -r 666w3wl4d
*copy output list of files to sublime text 3; select all lines; add to each:
gitk --follow [filename] > src/.notes/herewegoagain/[filename].diff
*examines....
Me: It's -10% done. you'll know I'm almost done when I enter the fugue state. You'll find me at this address. Give me this USB stick and a 4 pack of redbull and I'll do the merge.

When you accidentally type "git diff" in slack.

Git: Before you pull you need to commit your changes

*Doesn't even run a git diff to see what changes git was talking about.*

git checkout .
git clean -i
git pull

fck you visual studio!!! seriously what is wrong with you?!?
~me peacfully writing some code ~
ok let's see what we did
vs: I can't compile that. The key whateverKey in line 15 is not defined.
me: ok let's investigate...
nowhere in line 15 use whateverKey.... ok....
wait I didn't change that file at all.
~me clicking rebuild solution~
vs: can't build that because of whateverKey in line 15.
me : WTF?!?
checking git diff -> file not changed
me okkkkkkk......
closing visual studio and reopening solution.
Build succeeded.
What the actuall hell?!?
I'm spending way too much time trying to get that shity peace of software to do what it is supposed to do!

side by side git diff comparison 4 life

fuck inline

My biggest problem with Visual Studio Code is that every fucking piece of shit dev thinks it's their duty to introduce it to me. STOP. Just stop this shit, alright? Wanna use vscode? Fine, just don't tell me it's the best tool and I MUST use it instead of the tools I'm used to. I'm tired of this bullshit.
Every new project, every new team. Starting from js/java/.net monke and ending with PMs, I must hear this bullshit about god blessed IDE that I must use, because "why you need intellij/webstorm/rider? just install vscode and some plugins. we all use it in our project and it's ok".
FUCK YOU! Refactoring is not just renaming variables and extracting blocks of code into functions. If you want terminal integrated into your text editor with highlighting and LSP support, so be it. I want an IDE with rich refactoring tools, code analysis and good completion, database viewing/modeling support, good build tools support, good UI for git and git-diff, good test and code coverage support. I don't want your semi-IDE, bloated with hundreds of bugged third-party plugins, which I must spend a week on to configure and merry with each other before using.
JUST STOP this crap and let people use the tools they are proficient/comfortable/productive with.

I think the reason why git beginners have a hard time with it is because the api is a bit untuitive.

For example: if you want to "unstage" staged changes, you run git reset, and if you want to "delete" those changes from your working copy, you git checkout those files.

But then, you find out that you can do all of that if you git add . and git reset --hard.
So you're like "huh..."

And then you discover that if you end the resethard with a branch name/commit id then you also make current branch point to the commit or that branch/commit (respectively).
So you're like "huh..."

And also if you add a commit id or branch name to git checkout, you change the current branch to specified/enter detached state with HEAD pointing to that commit (respectively).

Oh and you don't use git branch to create branches, you use git checkout -b because it's a lot shorter.

So here's a rundown: git reset mutates things related to files, but also mutates things related to branches.
git checkout also mutates things related to files and mutates things related to branches too (in a diff way). Also, creates new branches.

I don't think this is intuitive. We users use the same commands for different purposes with just a different flag.
Commands shouldn't mutate different types of things. But don't composite commands (as in, "smart" commands that mutate different things) shoudln't be a flag in an existing command, it should be a single new command of its own.

Maybe if I reread the internals of git now, I'll be able to disgest the dozens of technical terms they throw at you (they are many). And in my mind, the api will cognitively fit to the explanations.

Here's another one that feels weird too.
If you want to make your changes start on top of someone else's commit, you do git rebase.
But git rebase -i can be used for that, and also to delete, modify changes or message of, reorder or combine previous commits of the current branch.

Maybe the reason why several things we do overlap with the same commands is because they internally do similar things, and while not separating those commands might make it less intuitive, it makes them more sensible? i dunno...

disclaimer: I'm not setting this opinion in stone though, and am aware that git was created by one of the most infuential programmers.

So, today I wanted to program a bit and, after reading the last chapter, I want to see what I able to do.
I run my last Linux distro, I open sublime and I start typing code. I finish, I build. 0 warning, 0 errors. Nice! I execute the code: error.
I watch and I struggle on the code for hours, I search on Google, I search on StackOverflow, but after 1 hour I notice I'm looking for a needle in a haystack. So I search instead for a way to produce a better error. I found it, I'm very happy. Let's try what the error actually is:
Error: success
Ok....
Ok...... Well, maybe.... Uhm......
Ok, I won't give up. I search for a tutorial. Found.
The code is almost the mine, it's actually a usual snippet, nothing new. I compare my code with the code in the example/tutorial.
First line, is the same.
First 10 lines, are the same.
First 30 lines, are the same.
I build and execute the example: it works.
I build and execute my code: still doesn't work.

I won't give up, I said it. I won't give up.

I wonder if there's a tool like git diff, so I can see what the differences are, maybe I've no good eyes.
I search, first Google result, "diff"
diff myCode.c example.c
"the files are not identical"

...thank you
I search for a better command
diff -y myCode.c example.c
"the files are not identical"

I search for a still better command
Found. StackOverflow stroke again.
sdiff myCode.c example.c
"the files are not identical"

.....
....

.....

I gave up.

Ps. I've 10 years of experience in programming

which *git* diff algorithm is your default and why?

## default (myers)
The basic greedy diff algorithm. Currently, this is the default. if you don't know whart you are using, this is it (and it sucks)

## minimal
Spend extra time to make sure the smallest possible diff is produced.

##patience
Use "patience diff" algorithm when generating patches.

## histogram
This algorithm extends the patience algorithm to "support low-occurrence common elements".

1. Find a function: getDayDiff(d1, d2)
2. d1 and d2 are momentjs dates.
3. See that function performs complex ancient math rituals and then returns an integer
4. Try to rewrite function, return d2.diff(d1, 'days')
5. Should be OK right? Run tests
6. Whole module melts down. WTF?!

Turns out the math performed returned the difference + 1 because it included the current day which moment's diff() function does not (out of the box).

Processes that depended on this function then uses the result like this:

const diff = getDayDiff(d1, d2)
if (diff-1 == should_match) { /* more fun logic */ }

$ git checkout .
$ run-shutdown-script-because-fuck-you

Gah I'm so mad at openCV right now. Is so stupid. Why the heck does it's kalman filter implementation accept 'float' as an acceptable type but not 'np.float32' gah gah gah it's so dumb spent half my day on that gah thank God for git diff it saved me cause it worked before and now gahhhhh. Anyway.

my git diff too big such that gitlab doesn't display it be default

oof

not a good luck when putting your changes out there

When you havent pushed in ages and have to tediously 'git status' followed by 'git diff <every fucking file>' to figure out what the hell you've even been doing since the last push and can maybe figure out another commit message than "Various bugfixes"

Just a friendly note: If you use the same technique as I i.e. local GIT repositories backed up on cloud, ALWAYS make sure there is still at least as much free space as the largest repo shallow clone.

It may happen that some really really archaic "commits" will disappear and git won't work as usual, but mostly you'll have an access to the branches, which you can diff against the master and make patches. Then just clone, make branches from patches and code happily ever after.

Also... you really really shouldn't push when you discover(git fsck --all) such an issue in your local repo, although I think git won't even allow you to do so.

I prefer VSCode over visual studio for a ton of tiny conveniences. Some git operations can really only be done conveniently by switching rapidly between the command line and the big scrollable list of diffs. The currently open file is automatically focused in the tree, not by explicit user command. Ctrl+Tab shows the last viewed section of files and not their name, so I can find an arbitrary point in my jump chain. If I open the diff for a newly added file it's possible that I want to edit the file, but it's also possible that I didn't notice that it's newly added. Painting the whole background green doesn't hinder the first scenario nearly as much as it solves the second, in contrast to VS not showing any changes, which just has me confused because of the total lack of modification marks.

I might sound stupid, but why don't solo-developers use things like dropbox for active file history that keeps track of every change and also gives diff options and up to 1 year log - instead of git?

Don't get me wrong, git is amazing when you have to work with a team or multiple people in general, but it's simply a pain in the ass when working alone and having to keep track of every state yourself.

Coworker just showed me how he avoids merge conflicts and I'm undecided on it. We use feature-branch workflow, so if a feature takes a long time to finish, it may mean merging master multiple times. He avoids it by stashing changes instead of committing them, then when he needs to merge master into the branch it's still clean. When the feature is done and he's ready to commit, he pops the changes and git diff shows all the changes before you push and you just change what you need instead of being forced to use the horrible merge software.

There must be problems with this, right? This seems too easy for it not to be the standard.

When a company makes changes to their Privacy Policy, they just have to provide the diff of the agreement.

The horror when you update team code after a vacation \°o°/

me: <checking diff of code, types "diff">
comp: error in command provide filenames
me: wtf...
me: oh, "git diff"
me: What's the diff?

Could a diff diff if a diff did not diff code? What?

Learn what blobs, trees, hashes and the concept of "commiting" (transactions) are.

Then learn how to use diff/patch.

Then learn Git.

In that order.

Want to open all the files changed in a PR?

> code $(git diff master --name-only)

#git #vscode #vscodehacks

If only git diff could be used to track changes in real life...

git merge "conflicts"

(not really an issue, but still a waste of time and concentration, when every now and then, using more than one branch, small edits, merge, and rebase, git diff complains about "conflicts" that are obvious to solve for a human but still not for a machine, despite the hype about the age of AI, coding co-pilots and the like...)