I have been using Git a long, long time. I have worked on Git clients and libraries. At some places I've worked, I am the person folks go to when they need Git help.
And yet, only today I learned you can pass -m to commit twice (or more) and it will do the right thing of making each successive message a new paragraph (which is useful for the convention of a short summary as a single first line and following paragraphs as a more detailed message).
@halfogre what!?!
Now Iโm even more mad at all the times I muscle-memoried โ:q!โ after a meticulous commit message 
@halfogre I once somehow managed to create a local branch called "origin/master"
That was fun.
@halfogre Same - worked a bit with git, and since learning it at uni, at every job since, I got rep of "git magician" for some reason. And while I am impostor-syndromeing about it, I learned recently a fun and useful thing about it:
Instead of push ... --force or -f you can also just use +branch-name in the command to convey that you want to force-push to remote branch with the same name. I'm still not using it often, but it's easier to write when I do it. 
@megmac
I used to use -m flags but eventually switched to using an editor.
Why? Because most of my projects are solo and I don't want to use an issue tracker. Instead I prefer to provide the context right next to the changes. Using an editor makes me use the description field, too.
It's a bummer that GitHub throws away most of the conversation in a PR unless you remember to update the first message before merging.
You see, I want to keep the information offline (because Germany).
@halfogre do you find using -m twice (or more) easier than entering a multi-line string into the shell? (bash, zsh, etc will prompt for additional lines of quoted string if you just leave the quote open; I do that all the time on commit messages.)
(The main hassle with โjust have a many line shell stringโ is you have to be careful not to prematurely close the quote by accident: which can require juggling โ and โ for quoting.)
> do you find using -m twice (or more) easier than entering a multi-line string into the shell
Yes, I do for my typical flow of one really short line followed by a still relatively short line. For longer stuff I just go to the editor.
I have never really done the open quote thing intentionally. Never felt natural to me. Editor feels more natural.
Lots of ways to do this, all just fine. I was just really surprised to find something new I didn't know about `commit`.
@halfogre Iโve gotten every used to multi line quotes in the shell (especially for awk / perl one liners), so it does feel natural to me. But I can see how itโd feel weird without 20 years doing that :-)
Thanks for the explanation. Definitely useful to know about repeated -m doing the right thing. Even if I suspect Iโd be more likely to use it in scripting than interactively ๐ค
@halfogre woah.
I've also been a very long time git user and this was news to me as well. Amazing.
@halfogre Looks like it's part of the spec. Not a fun read, but probably worth it.
Did not know that either. Thanks!๐@halfogre WHAT.
I use -m all the time!
@halfogre
I also did not know this, despite having worked with Git for the last 9 years or so.
Git, unfortunately, is what happens when developers think about functionality without thinking carefully about usability. It's probably an unpopular opinion on my part, but I'm not a n00b, I've used it a long time, and I still come to this conclusion. It's really capable, but it's truly terrible to work with.
@spatula @halfogre @feoh Git is popular but very few people actually know how to use it. It is especially facepalm-worthy when the โgit is the bestโ-types blow away their clone because they donโt know how to fix whatever they did.
FWIW, for all my personal projects I use Mercurial. It is so much saner you use - and the evolve extension is amazing. Going back to git feels like visiting the Stone Age.
@jeffpc Well, yes, that, and people who think that human beings manually moving pointers around is a good use of their time.
I've worked in this industry for 27 years, and I've never once rebalanced a binary tree.
@spatula Whenever I ask something along those lines, the I'm looking for a "I'd use a library or whatever code already exists that is well exercised/tested".
Similarly, when I'm the candidate, that's my answer to questions like that. I've *always* had the interviewer to agree with that statement. After that, the actual balancing feels like a formality. Granted, I still try to get the question right and while I'm writing the code and try to say what I'm thinking...but pointers don't scare me.
@jeffpc Yeah, pointers haven't scared me since 1994, but I am very much opposed to wheel reinvention, especially where very tried and true libraries are concerned. The chances of my making some weird corner-case mistake that someone else has already made AND fixed in their library are quite high.
It's all stuff I certainly CAN deal with, but it's not the class of problem I generally NEED to solve, especially when someone else already has done so on my behalf.
@jeffpc My workaround to this problem has been to leave the coding questions to other people and do more abstract interviews instead. One of my favorite things to do in an interview is to find some technology referenced on the candidates resume that I know nothing about myself, and ask them to explain it to me. It tells me a hell of a lot about a person.
Long ago, I used to like to ask people to randomize a list without repeating any values. Bonus points for knowing abt Collections.shuffle().
@jeffpc @spatula @halfogre Git's an incredible tool but it has the unfortunate feature of having some of the absolute worst human interfaces ever. It's an interface only a mad genius programmer with god awful social skills like Linus Torvalds could conceive :)
Because of Linux, and then Github, it gained a massive market adoption advantage.
One day, either Git will evolve to have a better human interface, or be replaced by something better. Until then, I'm glad that you're at least walking the walk and running Mercurial at home. A whole lot of people seem to just like to belly ache :)
@feoh The same thing can be said of that godawful turd of `systemd`. Yet another powerfully capable system with an absolutely bonkers and moronic user interface.
But somewhat puzzling is how they wanted to break apart `initd` which had become such a horrid monolith, and in its place, they erected an even BIGGER monolith, except with modules.
As a developer and CS person, I of course CAN work with obtuse systems like these, but that doesn't make it any less annoying to have to do so.
@spatula The other big criticism I hear from my crusty old sysadmin friends is that it violates long held contracts that the old school init system used to obey.
Please don't ask me for details, I'd need to invest time in dredging them up :)
Drew Miller
Bruvik 
Jake Carpenter
Arne Brasseur
Jari Komppa ๐ซ๐ฎ
jonny
Hazardius ๐ก๐ก๐ก ๐ณ๏ธโ๐๐ณ๏ธโโง๏ธ
AlinaX
Saphira Lohikรครคrmekettunen 
Adam Spiers ๐ช๐บ๐ฌ๐ง๐ฎ๐ช๐ฆ๐
Anders Marzi Tornblad
hauleth ๐ต๐ฑโค๏ธ๐บ๐ฆ
Kevin P. Fleming
Roman Luลกtrik
Andy 
matte
barefootstache
emily dogmom
Alex Hyett
Claudius (legacy account)
Michael
thanks for sharing! 
N-te
keithamus
fneddy
Nicolas Rinaudo
Meg
Ryuno-Ki
Ewen McNeill
Ryan Anderson
Brandon Ferguson
webhat
Jaycie
Max
Jeff Byrnes
That Blair Guy
hrbrmstr
ajfriesen
Don Melton
Kenneth Finnegan
Rory
Nicklas Johnson ๐ณ๏ธโ๐
Josef 'Jeff' Sipek
Feoh