You can reference the commits by SHA or relative to HEAD.

How is it ridiculously easier?

I assumed he didn't mean "git diff <commit_a> <commit_b>" but I almost asked anyway

I know how to do a git diff but that involves tediously finding and copy and pasting giant commit hashes after a git log and then hitting space x number of times to hopefully find what you want as opposed to clicking on the file. Checking out a commit in sourcetree is double clicking as opposed to more copy paste of hashes.

gui is way better for merge conflicts and viewing commits/prs (use GitLens).

also, I like committing in a single key stroke rather than add . then commit -m "blahblahblah"

command line is better for log, checking out whatever old commits or branches, especially stash and rebase. also undoing multiple commits.

I believe the git command you're looking for is git cherry-pick Suzzer. I use it all the time but not in the way you're mentioning.

I force pushed to master for the first time yesterday. It seemed to have the desired effect.

A bunch of large binary files were accidentally committed to master, making the remote unworkable. The bad commit is now headless, so should get killed by GC. Although it seems to kill GC rather than vice versa.

Sourcetree is just a front-end for CLI git, so there must be commands for everything it can do, it's just a pain to remember all the commands options and paths to files.

Also by using Sourcetree, you're not giving up CLI. You can swap between the two easily and type or mouse when you need to.

Something that made me lol hard in last week's Hacker Newsletter was lazygit, for this tagline:

1. You don't need to ever mess around with the long git hash. Look at `git log --oneline`. Better, `git log --decorate --pretty --oneline` is what I use. You can get away with the 7 character one.

2. You can specify the commits relative to HEAD, e.g. `git diff HEAD~5 HEAD~3`. Use an alias to abbreviate this.

3. If the old commit is a branch, you can use name the branch, e.g. `git diff master HEAD~3`. Use tab complete on branch names.

4. New commit defaults to HEAD, so you don't have to specify it if that's what you want, e.g. `git diff master HEAD` is the same as `git diff master`

But as I already mentioned to suzzer, I look at big diffs (e.g. merge requests) in Gitlab, and side-by-side, which is not something the cli can do natively to my knowledge.

Use `git commit -am`

or in my case, git commit -anm..

What exactly does -n do?

No verify i.e. skip pre-commit git hooks. In my case its a (self inflicted) lint:all hook.

ugh you have commit hooks? that seems like overkill. we have push hooks which is annoying. seems that only running hooks on merge is best imo.

You can have very basic commit hooks. We have a hook that adds a change-id for Gerrit.

Pre-commit must pass lint-all. Pre-push must pass all unit tests. I made it and love them . Though I told my team to feel free to use git commit -n and git push --no-verify on their own branches prior to doing a pr. Haven't had any complaints.

I like that system.

Since we're talking about git and aliases in general, here is part of what my current one looks like. I have some for git diff of various branches (master, main dev branch, etc) but those didn't fit in first page.

pr hook should be unit tests, lint and build. Then no-verify when addressing pr feedback and whatever. Then run hooks on merge.

We don't do that and so many devs waste a lot of time on pushing and amending prs.

Also our hooks take like 20mins.

Imo commits are personal. I mean, you gotta commit to change branches. Crazy to force that.

we have one that runs on every commit to master

runs unit tests, blackbox tests, and some performance regression tests.

but sometimes even the jobs that run on git hooks is bugged. This week I found out one of our jenkins jobs was silently failing, probably for weeks now. I told the girl that wrote the script, she blamed the infra guy, the infra guy blamed her, and my boss lost his ****.

Ended up being a problem with using a deprecated linux command.

i have a question for you guys which will sound totally stupid probably -

How does your guys' work schedule work? Do you need to request days off? Do you get a certain amount per year? Are your work hours set, or do you come and go day to day as you please?

My current job has "unlimited vacation time" which really means "most people take less days than if we just gave them 3 weeks." I give 2 weeks heads up for anything longer than 3 or so days, if I wanted to take tomorrow off I'd just IM my boss. For the most part I tell them when I'll be off, I don't have to ask. I have been politely asked to change dates once or twice in my career.

80% of my jobs have been fixed amount of vacation time, usually about 3 weeks.

Work hours are more or less at my discretion but there's a spoken rule that there needs to be significant overlap in hours so anyone whos' working will typically be working during 10am-3pm. But today for example I came in at about 10am, yesterday I started work at 7am.

All 3 companies I have been at have had a set amount of days (both PTO and sick days which has always been an honor system to use), and a system for logging them. My current company switched to unlimited, and I don't see anyone using less tbh. In fact most of us plan to use 20ish days, whereas it was 15 prior.

Same thing with work hours as Rusty. Although it is kind of unspoken to be in before 10 at my current job and it was directly set as you have to be in the office between 9:30 and 4 at my first.

My 2nd job was insanity. We had people that came in at 6am and people that came in at 2pm.

All 3 were fairly opposed to remote work, though I do 1-2 days remote currently because my manager is into remote work

I had a boss who was opposed to WFH but didn't want to outlaw it, so he'd just schedule meetings during when I would usually WFH. (He knew I did not like to attend meetings remotely if I could avoid it)