Hacker News Reader

Comment by jakub_g | Hacker News Reader

Slightly meta: Turning off the syntax highlighting entirely seems a bit hardcore to me, however, I always read each commit I do via git diff / pull request view, which does not have syntax highlighting (other than red/green for removals/additions), and I always strive to make the diff look good for each commit (which means, doing one thing at a time in each commit, and making sure it reads well).

Unfortunately, not everyone I work with reads their code (yep, they write-commit-push-and-forget!) which sometimes makes code review a not so nice experience.

Please, please, read your diffs before sending code for review.

View on HN

I agree completely. On that note, I also consider `git add -A` to be horrible practice. Adding files manually makes you tons more aware of what changes you actually made, and also prevents you from accidentally adding extra files you don't want.

I also recommend people to look at the Linux kernel git history for good examples of how to use git well. Even if you have no knowledge of kernel code (and most people don't) it's still extremely clear how each commit makes a single change and is fairly easy to review. Now, the Linux kernel has the luxury of being able to demand people fix their commits, but that's not to say you shouldn't strive to reach similar states in your commits.

    > I also consider `git add -A` to be horrible
    > practice. Adding files manually makes you tons more
    > aware

Agreed, but I take it further than that - I always use `git add -p` because it forces me to read through everything, and make everything in the commit deliberate.

Even if you think you just made a slight change to a file or two and definitely want everything, you'd be surprised how often it lets me pick up some aesthetic change, comment change, or dependency version bump that's completely unrelated to the commit message I'm about to write.

I am in the other extreme: I always use `git add -A` (but I read the diff before) and avoid doing more than one thing in the first place (which would make me want to use `git add -p`). IMO it is an antipattern to do bazillion changes at once, and then do several commits out of it (unless you really know what you're doing) - because you commit a state of the repo that (probably) never existed. I prefer to do change, commit, do change, commit, and at the end do interactive rebase and perhaps squash some things together.

> because you commit a state of the repo that (probably) never existed

The state doesn't exist in the repository until the code has been committed - so your statement is a little backwards. Sure the code might never have only had that edit in it, but the repository has no changes until a commit is made.

What the commenter means is that one might accidentally select some invalid or non-sensical set of partial changes (that won't compile or run) in an effort to create a logical progression between state A, "no feature", and state B, "feature is complete".

If a bug is found, the invalid state of the intermediate commits might make it harder to isolate the change that caused the bug.

That being said, I generally use "git add -p". I've gotten accustomed to trying to have clean commits. Although, sometimes the changes have enough complexity that breaking them up becomes too painful and I end up with a few messy commits. Such is life.

Don't you always have commits chains like "add debug printfs" "fix bug" "remove debug printfs"?

That's a scenario for branch-then-cherrypick :)

Branch, write test to expose bug, add debug aids, fix bug, switch branch, cherry pick the test & fix, delete debug branch.

Or just `git rebase -i origin/master` and delete the unnecessary commits without additional branching.

I have these in my WIP branches. When I’m ready to merge, I squash out some of the smaller commits, and then merge without fast-forward. Reduces noise and keeps a clear history of groups of work, at the expense of very cheap branching and local rebasing operations.

    > and avoid doing more than one thing in the first
    > place (which would make me want to use `git add -p`)

Yep, I still _avoid_ it, but I'm often surprised by something that slips through - e.g. a typo in a comment that I just changed as I read through - but a comment somewhere totally unrelated to the present commit.

Maybe your better at avoiding it than I am, but I find it to be easily done.

If I do such minor corrections, I usually commit them immediately (via `git add -p` or better `git commit -p`), so they are out of my way.

I understand what you're getting at, but unfortunately I just don't write code that way (And I'm not really sure I know tons of people that do). I always end-up making changes or realizing things halfway into working on something. That said I do understand your point, generally what I do is commit 'logical' parts of my work (Generally using `git add -i` to commit individual chunks), and then when I'm done I checkout each commit to test it compiles on its own.

The fact that `git` makes changing your history around easy make this workflow work fairly well without too much extra overhead to achieve a clean history. I don't disagree that doing it right the first time is really the best way to do it, I just don't find that it really works out that way in practice.

True, sometimes I realize I need to do something half-way when doing something else, then I do a "WIP commit", do another thing and commit that thing (sometimes in a separate branch forked off master), then perhaps reorder commits if needed, and go back to continue working on the previous thing (and amend the "WIP commit" once done)

The tricky thing is that when doing several things without committing, sometimes you can end up in a state of the repo where you can't split the work done into nice commits anymore (because you might have modified same part of code multiple times).

I usually have a paper and pen at hand, all non-current work goes there — empty circle and todo text. It was my bad habit to patch things immediately at sight. That way I sometimes missed part of context, either of main task, or secondary, depending on complexity of both.

Maybe this step-by-step working scheme was influenced by svn that cannot commit partial changes, but it seems a good practice on its own, especially in total refactoring.

  git status
  ...

Yep, those are the things I changed

  git add -A
  git status

Yep...

  git commit

...editing in VI... :wq

Mischief managed.

Yep that's the way to do it. Assume that you want to add all files (which is usually the case) and then just double check before. This is my practice.

I think this is a good practice; it's always easy to let your attention lapse and skim over a line when you're reading the full diff. `git add -p` is nice in that it explicitly requires you to ack each change.

I recommend using `git commit -p`, which is a shorthand for `git add -p` followed by `git commit`.

Then, `git show` before you do `git push`.

I use `git add . -p` which allows me to 1) selectively create my commit, and 2) re-read what I've changed to understand it all better. I agree about wholesale adding, but of course there are exceptions when you already did `git diff` Or something.

Side note: `git status` is not a good means of "knowing what was changed". I see others use it constantly (doing git status -> git add . -> git commit) and it'a sloppy IMO.

I just don't understand why people use the git CLI. Git is too powerful a tool, using the CLI limits one of two things: your capabilities, or your productivity.

Often times I see my coworkers screwing around with git trying to review diffs and commit their changes. One of two things happens: they spend longer than they should doing it, or they do it haphazardly. Sometimes they get it wrong.

Just use a GUI.

I use SourceTree and often find myself committing individual lines or chunks in my files instead of the whole file. I'm confident the average git CLI user doesn't even know this is possible. I can do massively complicated commits with ease because I can see every single line that has changed right in front of me.

I just don't understand the fetishization of source control via CLI. In all likelihood it doesn't make you more productive, and it's probably less accurate unless you really know you way around git.

> I'm confident the average git CLI user doesn't even know this is possible.

Your sibling commenters disagree.

In that case, I'll support his conjecture by admitting that I didn't know that was possible. I'm a pretty average git user, if not actually using it slightly more than my peers (many of whom are only switching from subversion to git within the last 3-4 years). But the rate at which I become more proficient in git is extremely slow - you have to go out of your way to learn more about git most of the time. Maybe a GUI would help in that particular aspect, since I could see unknown options and experiment with them.

As someone in pretty much the same boat, try SourceTree or the like. I still use Git from the command line, but it really opened my eyes on what Git can do.

Agree with this 100%. If you can do everything you want on the CLI then more power to you, but in my experience people using the CLI for diffs/staging include unrelated code in their commits at a much higher rate. I checkout/branch/pull/push in the CLI, and use Git Tower to diff and stage.

As a counterpoint, every time I've tried a git GUI (I've tried lots), I end up bashing my head against it within a few minutes. They usually make very basic things easy and anything even slightly outside the box basically impossible.

I'm much more productive using a couple of git commands that I can type in a few seconds, rather than hunting for things in menus and dialogs for ages.

> committing individual lines or chunks in my files instead of the whole file

That's really git 101. I'd expect if you haven't learned to do that via the CLI then maybe you gave up too soon and jumped straight to GUI? `git add -p .` is very easy to use.

[GUIs] usually make very basic things easy and anything even slightly outside the box basically impossible.

Whereas in the CLI, complex operations are difficult but possible; trivial operations are also difficult but possible. :)

did you try magit?

Maybe it is of limited use for non-emacs users, but imho it's a fantastic way to use git with a perfect balance between making simple things very easy to do and allowing (almost) anything you can do with the CLI while providing excellent visual feedback.

I also found the git plugin [1] in the Sublime Text Package Control repository to be rather useful. The basic idea is similar to magit, with the (subjective) upside of avoiding the crazy Emacs hotkeys.

That said, since I discovered Spacemacs, I'm also enjoying using Magit - the "evil" control scheme for it is great, too.

[1]: https://packagecontrol.io/packages/Git

magit is a marvelous piece of work indeed, and the thing that keeps me going back to Emacs every time I get curious about some other editor.

But yeah, I picked it up after quite some time working in Emacs, and the keybindings feel natural. Perhaps it would be very different for somebody new to Emacs.

i use a gui for two things:

1. gitk is the handiest way to undo the occasional accidental commit ("reset index, leave working tree untouched"). yes, i know there's bound to be a CLI way; no, i can't be bothered to look it up and remember it when i have gitk right there and it's not a mainstream use case.

2. git-cola does interactive commits way better than the console interface does.

> I can do massively complicated commits with ease because I can see every single line that has changed right in front of me.

You mean `git diff`? I'm sure it comes down to a matter of preference, but claiming that using git from a command line slows you down or limits your capabilities only makes sense if you rarely touch a command line (a bit strange for a developer). Plain old git from the command line is functional and actually quite user friendly. Add your own aliases and some kind of git shell and you can fire off a complicated commit as fast as you can sensibly review your diff — all without lifting a finger from the keyboard.

GUI git tools are nice for visualizing feature branches, but for the basic day-to-day workflow the CLI is much more productive.

> Git is too powerful a tool, using the CLI limits one of two things: your capabilities, or your productivity.

Surely the exact comment, with the roles of 'CLI' and 'GUI' switched, could be made without anyone necessarily noticing. The anecdotal evidence of what works, and doesn't work, for you should not, I think, be regarded as any particular evidence of what does work for anyone else, much less what can work for everyone else. (As has already been pointed out, your child posts are full of responses from people with other experience.)

I would wager this goes beyond just `git`, and is more a statement about CLI vs. GUI, which IMO just comes down to preference, and how much you're really willing to learn one vs. the other.

There are definitely people out there (I've seen them) that would be better served with a GUI because they really can't be bothered to learn how to use the `git` CLI well (Or just use `git` well in general). But at the same time I'm tons more productive with the CLI vs. GUI.

Unfortunately, nome of those neat git GUIs work on Linux...

Not true: https://www.gitkraken.com/download

Propietary, requires an account and mergetool is behind a paywall (although this is not as significant). Ouch.

SmartGit is wonderful - I've used it for years - and it works on Linux, macOS, and Windows.

http://www.syntevo.com/smartgit/

If you use Emacs, check out magit. It's not really a GUI but it is git integrated into emacs so you can do all the usual functions, view diffs, write commit messages, etc. all from within emacs.

I agree with this 100% when it comes to creating commits. It makes it easy to create sensible commits with only the changes relevant to the individual commit. You can see what's staged and tailor it quickly. I'm a heavy terminal users and I agree that in general committing from the CLI seems like using the wrong tool for the job.

That said, there's a lot you can't do easily or at all from the GUI. Rebases, surgically reverting a series of commits from a bad merge, re-writing branches, blame, bisect, reviewing the history of a single file, checkout of a specific version of a file from a previous revision; there are a lot of things that I can do easily from the CLI that I don't know how to do or don't care to do from a GUI.

Use the right tool for the job. The GUI is great for common tasks and the CLI is great when you need more power.

I used to be in your camp before I discovered git add -p. Allows you to interactively pick which hunks to stage. SourceTree essentially replicated the same workflow in a nice GUI.

No gui that I have seen can deal with the GIT workflows in an efficient manner. Most GUIs are slow. Rethink GIT workflows and make them worse. I only use a GUI for this: - GitX for adding and reviewing files to a commit. (partial commit support is also very nice).

the rest is command line git.

With command line completion is pretty good and fast.

> I just don't understand why people use the git CLI. Git is too powerful a tool, using the CLI limits one of two things: your capabilities, or your productivity.

That's fine, but on the other hand, a typical GUI limits both your capabilities and your productivity.

About the only interface I prefer over git CLI is magit in Emacs. It's unlike your typical GUI, because you don't have to click at anything (keyboard-operated), and it exposes like 99% of git inside a well-thought-out text-based UI that adds just the right amount of interaction and editor integration. Using magit is incredibly efficient, much faster than even using CLI git.

> I just don't understand the fetishization of source control via CLI. In all likelihood it doesn't make you more productive, and it's probably less accurate unless you really know you way around git.

If you don't know your way around git, chances are you're going to seriously screw things up via GUI and you'll have no idea what happened or how to fix it. Starting from CLI is a good way to actually learn the tool, which I believe is crucial regardless of what interface you'll eventually be using.

Yeah that's what I used to think but you have at understand that typing and keeping your fingers on the keyboard is often way faster than moving around and clicking various regions of the screen.

Then what happens when they update their GUI? Oh you have to learn it all over again.

That being said, I do on occasion open a GUI if I have to do a complicated merge or something like that but for basic operations why would you need it.

I always do `git add -p` and `git commit -e -v` so that I know exactly what's in the commit

> On that note, I also consider `git add -A` to be horrible practice. Adding files manually makes you tons more aware of what changes you actually made, and also prevents you from accidentally adding extra files you don't want.

I use `git commit -a` a lot, but only after checking `git status` (or `git diff`), which is probably my most-used git command.

Those two A's have completely different meanings though (add all, commit amend), not sure I understand how it's relevant to the parent?

Grandparent is correct: "git commit -a" is --all, not --amend.

I get a lot of (good-natured) crap from other developers because I always use a graphical tool to view my git diffs and make commits. I realize that there are console tools like git interactive staging that provide similar functionality, but honestly I just like having a dedicated windowed application always running that is always the gateway for my code.

> I get a lot of (good-natured) crap from other developers because I always use a graphical tool to [...]

One of my colleagues uses PyCharm, while at the same time using Vim somewhere else (for both Python and not Python). This frustrated me a little every time he asked me to his terminal in a similar fashion I'm annoyed by somebody using GUI for the simplest task I could do in a single command. In both cases I try to keep comments to myself, at least now that I'm a little wiser than I was five-to-ten years ago.

It takes a lot of patience not to throw such "good-natured crap" if one is proficient with more-basic-yet-powerful tools. It needs wisdom to recognize that people do things their own way because it's more effective for them at the moment, and it's hard to change one's habits on a whim at arbitrary times. It gives me much better results to call others to my terminal to show them some interesting thing; they are in much better condition to learn a new Vim command or CLI tool (this way I successfully exposed two colleagues to AWK, what I planned for a long time).

I'm amazed someone wouldn't read their diffs, just for their own sanity. It's a great summary to remind yourself what you just did to resolve an issue. And what if your cat just walked over your keyboard while you weren't looking? Your diff will tell you this. Silly but meaningful example.

>Slightly meta: Turning off the syntax highlighting entirely seems a bit hardcore to me, however, I always read each commit I do via git diff / pull request view, which does not have syntax highlighting (other than red/green for removals/additions), and I always strive to make the diff look good for each commit (which means, doing one thing at a time in each commit)

...alas, some projects require 1 commit per pr, max...

...then again, if you do a 1:1 commit:change as default the rebase msg is much easier to structure :)

I prefer doing this with 'git gui'. It's standard and is very simple to use. What is important you can easily select chunks or lines to stage or unstage.

Learning to this both after writing the code and being able to plan what chunks of works I do when is a game changer and really useful when you work on a team.

I sometimes find waiting until I am done working on something to break it up into small commits can be a headache.

Agreed. The first thing the older guys taught me at my first internship was `git add --patch`.

I always use `git commit -v` (with alias).

You get the diff in the editor when you write commit message.

I'm just trying to convince folks to actually write commit messages. :) I don't think anyone is relying on diffs to find bugs. Do they?

Similarly, before doing something complicated at work, please talk about it with a coworker first. Last thing I want is to try to decipher a complicated piece of code without a good motivator as to why I'm doing so. Advanced warning is a great motivator.

I do this, but not mainly to find bugs, though on rare occasions I do. I almost always find something, e.g. stylistic aberrations, overzealous changes from an intellij refactor, debug statements I forgot to remove, part of another change I'd thought I'd stashed but somehow ended up in the commit. It blows my mind that anybody would not read their diffs.

Fair. I try to do this. I still make mistakes, I'm sure. This ultimately comes down to a "good intentions." Would be neat to make a mechanism out of it.

Reply on news.ycombinator.com