Git Overview - Computerphile

[@kbreezy004]

When British folks type `git init` does the shell respond with "yes"?

[@at0mly]

lolol

[@rmsgrey]

Nah, mate. You need a double 'n' there, innit?

[@blackenedsprite8542]

In London yes, in any other part of the country it just initializes the repo.

[@Knightfall23]

😂😂😂😂

[@nils-erikolsson3539]

😂😂😂😂

[@teppichboden2]

It's unironically one of the best pieces of software out there. It's super useful, free and open source, and incredibly efficient.

[@GrimmerPl]

And it all started with one guy being angry about software versioning option available at that time.

[@pandaDotDragon]

and also super counter intuitive. A sh*t load of tutorials and still nobody understands how to use git with the command line (except the same 4/5 commands).

[@morosis82]

@@pandaDotDragon nonsense. I started with git in 2008/2009 I think, back then you had to really understand it to use it, but the modern changes to the command line make using it a lot easier. It's much more intuitive now, but you still need to understand what you're trying to do, and it provides access to the plumbing if you need it. Filter branch was a godsend when someone accidentally checked in an auth token or some giant binary blob.

[@pandaDotDragon]

@@morosis82 Well I used to work with a committer to the git project and he was well aware of this issue so... That's why a lot of people preferred mercurial: the command line was easy to understand, easy to use.

[@morosis82]

@@pandaDotDragon I think it's a lot of just people are unfamiliar with using a command line on a daily basis in general. As someone who learned git early on, and has been a Linux daily driver for over a decade, mostly the GUI tools (apart from merge/diff tools) just get in my way. When you know what you want to do, cli is just better/faster.

[@at0mly]

decent collaborative versioning systems have been one of the most important advances in software engineering. i say this as somebody that started programming over 20 years ago when we were literally emailing source files and tarballs around, before we got cvs (and then subversion and so on)

[@johan.mp4]

We had version control back then too, so no need to email files :)

[@markostojanovic6973]

@@johan.mp4 true, but accessibility of different servers and syncing was a pain. e-mail and sending diffs was often faster (but not reliable for the history)

[@johan.mp4]

@@markostojanovic6973 Git is for sure more convenient. But SourceSafe, which I used when I started out, worked pretty well too. I believe it was released in the early 90's. There are of course earlier systems than that. What I remember, was that branching was more complicated. But syncing worked well.

[@at0mly]

maybe at large companies with the resources to run it and willing to pay for the software and the staff to administer it. even then it was generally one central server and terrible with any complicated branching.

[@johan.mp4]

@@at0mly we were a three people company. We realized pretty soon that we needed a VCS. Not only for collaboration, but also for backup of the code. I would probably have used it on my own as well.

[@Android480]

I’m embarrassed to admit I’ve been using git professionally for years now, and really don’t know how it works internally. I have an idea about how it might track differences rather than just raw copies, but it’s magic to me.

[@FabioRossettiFI]

it stores snapshots, not diffs - other source control softwares do, though

[@p0n-pompf]

Isn't it like this with everything related to computers though? Even machine code is just a higher level language to the inner workings of a processor in a sense - as far as you know the input, you know the output it's everything that matters in the end

[@tackline]

I still don't know the algorithm diff uses.

[@menachemsalomon]

@@FabioRossettiFI I believe in the compressed (packed) form, Git does in fact store deltas, which are based on diffs. But they're on chunks rather than on files, and they can be reused, somehow.

[@wizard7314]

@@menachemsalomon if true, same thing that backup programs do, like duplicacy or restic.

[@TheMilli]

I really would like a full video on how the .git folder actually keeps track of my version history. That would be incredibly interesting!

[@shapelessed]

What would be incredibly interesting is an explanation on how thatovercomplicated, overengineered bloated software works.

[@TheMilli]

@@shapelessed I mean, that is exactly what I'm asking for, isn't it? Also, since it sounds like you don't like git, what would you recommend instead and why? 😊

[@maharjanlaxman]

True

[@yondaime500]

There's a talk called "Rules to Git By" in the Coding Tech channel that does that.

[@codycast]

There are 100s of KZbin videos on the topic.

[@MrTridac]

I like the fact that Linus Torvalds developed git mostly out of spite. But now - like Linux in the world of OS's - it seems to take over the world of code management.

[@lawrencedoliveiro9104]

If there was any “spite”, it was not on his part, but on that of Larry McVoy in cancelling their BitKeeper licence.

[@MrTridac]

@@lawrencedoliveiro9104 I don't know the exact backstory, but in an interview he basically said: "I was like f*&k BitKeeper it's s#!t anyway, I'll do it myself".

[@lawrencedoliveiro9104]

That was after the cancellation of the BitKeeper licence.

[@elliott8175]

@@MrTridac Sounds like Linus...

[@BruceHoult]

@@lawrencedoliveiro9104 don't forget Tridge's part in it...

[@Bob_Burton]

This would have been so much better with some demonstrations of Git in action

[@blackenedsprite8542]

Scott Hansleman has a fantastic video on git (possibly more than one 🤔). KZbin him :-)

[@AndersJackson]

No, I really don't think that live demo would have helped. This is about actually understand what happen without using the tool. Would ACTUALLY helped if doing git with ordinary shell tools, not git. But that would probably been a bit to advanced for this channel And there are other videos that just do that. And they are great, if you already know about git.

[@DaVince21]

No, I think explaining the theory first and demonstrating it in a part 2 was the better idea.

[@michaelcumming1233]

Git is magic. I used to use it daily and forgot how it worked. Git, I think is one of the foundations of our technical culture. Isn't it surprising how so few people know about it and how, if you mention it to anyone at a party, their eyes glaze over immediately?

[@badmanjones179]

>talking about git at a party yea why dont u bring up object oriented programming or maybe regex and see if that gets ppl more excited

[@AndersJackson]

@@badmanjones179 yes, regexp are fun. I would love talking to some people about regexps and even grammars at a party. To few such parties though.

[@lonelylongdistancekiller9844]

@@AndersJackson hey that's how you get invited to parties ey? I'll stop talking about functions and boleans and start talking about git

[@Kuchenklau]

I'm in my 5th semester of CS and while I've learned a whole lot about algorithms and Turing Machines, I'm an absolute git illiterate. So this video is much appreciated :D

[@richardaversa7128]

Illitergit

[@stef9019]

I was wondering if you guys have ever done a video on memoization, I think it'd be a great topic to cover for this channel, since you've already covered recursion, they go hand in hand.

[@crazychicken0378]

Tabulation would go hand in hand with their loops video as well

[@AndersJackson]

I think they have done that.

[@00ffdc]

Hey, nice electric upright bass at the background!

[@Desolator-fz5em]

I remember back in my college days where I just make an archive of my source code for thesis (a record management system), and I name the archive like "rms-v1.0" and when I put some changes in source code i copy the archive again, put the changes and rename the newly updated archive into v1.1 or v2.0 whatever. Just a manual versioning of my source code back in the days. 😂

[@IceMetalPunk]

I can't imagine doing my job without git (or, I suppose, some other versioning system). But I still remember the very first time I ever used it, back during a university group project. None of us had ever used git before, and while we were learning decently, we all panicked when there was a merge conflict *in a PNG file.* We didn't know about git reset, or git merge --abort, or even anything about what a merge marker was; all we knew was that we'd pulled some changes and suddenly our logo was completely corrupted and we didn't know why. That was a long night... To this day, I still think git needs better merge marking of non-text files...

[@lawrencedoliveiro9104]

How would you do it? With text files, you can narrow it down to lines which have changed, and show those lines to a human user. What would the equivalent be for binary files?

[@KimYoungUn69]

Lol how? U clearly do not knows how it works under the hood

[@jonathonthomas3048]

Git has no method of source control on non-text binary files. Changing one pixel on a PNG doesn't necessarily change only a single bit of data. Depending on what kind of compression is used and the file format used, two files with only miniscule differences between them could have drastically different binary data. Take into consideration not only PNGs, but JPEGs, GIFs, BIN files, compiled applications, ZIP files, etc. For git to be able to differentiate between two arbitrary binary files, it would need to know the format of that extension, how the data is arranged, any compression used, and more. This is way beyond the scope of what git is intended for, and all git can do with binary files is say, "hey, these are different" and leave it at that.

[@IceMetalPunk]

@@jonathonthomas3048 No... that's not entirely true. There are many image manipulation libraries, for instance. If Git knows that PNG 1 and PNG 2 are different, it's not hard to create a third PNG that's a side-by-side comparison of them. It doesn't need to highlight the exact parts that are different, just offer the two versions next to each other so the user can crop the one they want and commit it. Git doesn't need to know which pixels changed, or do anything on a bitwise or bytewise level; it just needs to use a standard library to concatenate two images together. But either way, as I said before, even without all this fanciness, having Git output a report that says "hey, these are different" for non-text formats instead of injecting text-based conflict markers into a binary file (and thus corrupting it) would still be better than the current behavior.

[@briancoverstone4042]

The real power of git is where it figures out when and which files change (especially when there are hundreds of thousands of files), what the differences are between any two versions, merging with all kinds of different conflict resolve mechanisms... source code becomes moldable like clay, so much more than any other source code system that I've used.

[@lawrencedoliveiro9104]

Remember that Torvalds himself describes Git as the “stu

[@lawrencedoliveiro9104]

pid content tracker”. He deliberately avoided very elaborate conflict-merging schemes, preferring to concentrate on narrowing down the points of conflict so the human user has just the info they need to resolve the issue, without needless distractions. He said he wouldn’t trust his code to more complex automatic mechanisms that many others were proposing. Yes, “clay” is exactly the word that has come to my mind, too. Having all your work safely stored away in the commit history gives you the confidence to take more risks, knowing that, if you screw up, you can always go back. ;)

[@paxdriver]

a live demo would've been peeeerfect in this video ;p thanks guys, love your work

[@AndersJackson]

Really don't think that. Not to this channels audience. There are lot of videos where you can follow how they demonstrate git tools. It WOULD be useful if the management would been done by ordinary shell tools, only use git status and git log to verify the work. There are still such videos out there.

[@paxdriver]

@@AndersJackson they do demos all the time on this channel. C, C++, Python, I've seen several low level code demos on this channel, a short 7 or 8 terminal commands wouldn't be out of the ordinary at all.

[@jasonzvaniga]

I wish you would say `files` instead of `codes` .. Git is useful for any files/folder, not just for coders/codes.

[@arngorf]

Except that binary files are not well supported, at least under basic git. Some other versioning systems, for instance, has support for displaying images, videos and 3D assets natively. I would love for git to improve for these kinds of use-cases.

[@retropaganda8442]

Some people even use it to store the evolution of french law, which incidentally, is also called "code" ;p

[@wizard7314]

That's not true. Git is really only useful for line-based files where the function mapping from semantics --> file content is continuous. I.e. small changes to the file give changes only in a few lines. It uses this feature of plaintext files (particularly code) to store diffs. Git is unusable for binary files or other files that don't satisfy these criteria. Every git commit you make stores the change to the files. So the changes better be small. With binaries, it nearly rewrites the entire file when you make a small change, at least from git's perspective. Especially if it is on one line, then it does entirely rewrite it. Now there are other software packages that even handle 'binary' diffs well. Backup software, like duplicacy or restic. There's also general compression algorithms for arbitrary files. A question I have is: is there a VCS that combines line-based and binary diffs?

[@brucewayne2480]

@@retropaganda8442 où est ce que je peux trouver ça 😛 ? C'est super intéressant

[@retropaganda8442]

@@wizard7314 AFAIR, in the most basic storage format, git doesn't store text diff, but an entire copy of the file. Last time I looked was probably git 0.x but I don't think the format ever changed since, except for Large-File-Storage. Now it also seems to sometimes consolidate files together into "packed refs" or something. Haven't had the curiosity to look into the details. It's probably pretty basic like the rest. In general, git becomes a problem with large repositories, and you have to fight the tool with the most obscure options to save disk space and network. Simplicity of the implementation comes with its costs. Unsurprisingly, it has far better performance on unices where you can work with medium-sized repos with no pain, but the performance on windows where the filesystem is slow makes it frustrating to develop on this platform.

[@videopsybeam7220]

I like to think of Git as an extremely high-level UNDO button.

[@Slarti]

I'm 6:45 in and I haven't heard the word "repository"...

[@alexesteh]

Someone said git is the only implementation of blockchain that works and isn't a scam

[@drdca8263]

That’s an overstatement. I think the better criticism of blockchain projects like bitcoin, ethereum, etc., is that their benefits aren’t worth the environmental costs (assuming PoW), and are also highly overstated, not that they “don’t work”.(“scam” is a kind of fuzzy word I suppose, but it also doesn’t seem quite appropriate to me.) The question of “are they actually used in ways that produce economic value?” both seems an interesting/important question, and also seems an ill-defined question. The only theory of economic value that has made much sense to me is the theory of subjective value, that economic value of something is determined by the value people assign to it, but, I do sometimes wonder if there is something which it is missing. But, what that something could be, I have difficulty describing to myself. It seems like some kinds of resources, are like, “actually useful” and “not something people are valuing for silly reasons”, but what does it mean for something to be “actually useful”? ‘ “Actually useful” *for what?* ’ I ask myself. The usefulness of a given resource seems like it would necessarily be in terms of what we would want to use it for? But, perhaps if we look at like, networks of what resources we can use to produce what other resources, and look at the inefficiencies of different cycles of conversions, there could be, like, some eigenvector decomposition junk, where looking at which eigenvalues (all less than 1) would be the largest, or something..? Like, suppose you could convert 1000 gallons of water and two cheese graters to 100 hamburgers and 2 cheese graters, and you could convert 100 hamburgers and 2 cheese graters to 900 gallons of water and 1 cheese grater, then, like, Well, going back and forth like that would convert 1000 gallons of water and 2 cheese graters into 900 gallons of water and 1 cheese grater. Erm... The idea I guess being that, perhaps we can separate the concepts of changes of who has which resources, from the concept of which resources exist, and then, well, we can’t make something out of nothing, so in some sense the amount of resources that exist and isn’t waste material should always be decreasing, but, if we neglect “the hydrogen the sun is burning” as a resource being used, and consider the light from the sun to be a continued source of resources from outside the system, and also I suppose we could consider resources extracted from the earth to be additions to the resources that are available, well, this still doesn’t seem to exactly account for “value”, but, if we looked at “what possible things could we do”, and like, only added a little bit of info about “what kinds of things do we care about whether we can do them?” maybe that could be a direction towards a more objective way to define “how wealthy are we as a society”? So, maybe using something like that as an additional extra thing to add on to the subjective theory of economic value? .... why did I start talking about the nature of value again? Something about cryptocurrencies? Ah, right. Yeah, like, when the prices go up, what does that mean? Well, it means that the price at which people are willing to buy and sell it goes up. And, if a person’s “material wealth” (I don’t mean this as a point against focusing on owning things. I’m just questioning what the best way is to evaluate “how much material wealth someone has”) is considered to be the dollar value of their possessions, or, how much other stuff they could buy at the current marginal prices, then this increase in price is an increase in their “material wealth”. But, what is this reflecting “in reality”? I suppose this ties into the question of “why do currencies have the value that they do?”. What is the “real meaning” of the exchange rate between Norwegian Kromer and the British pound changing over time? Like, yeah, at the basic level it means the amount of one you can exchange for the other is changing. But like, why? When it is changing, is the wealth of the person with money in the one currency increasing, or is the wealth of the person with the other currency decreasing? What’s going on? (In my town) All questions.

[@barneylaurance1865]

Git uses a merkle tree, but it isn't a blockchain because it doesn't have a consensus mechanism. If different people commit different things to their own copies of their project git leaves it up to each person to decide which commits to pull in and which to ignore. It works fine in a team where everyone works together and people can agree on what software they want to build, or in a open source project where the leaders manually pick and choose what contributes to accept and what to reject. A blockchain would have some automatic built in way to control who got to add commits and avoid creating multiple incompatible branches of the software, because it supposedly needs to solve the Byzantine Generals Problem,

[@4_real_bruh]

You can also use git to automate backups if you're sick of services like Windows Backup by creating a local repo, it doesn't *just* work for code, even if that is it's main purpose

[@phizc]

It's really inefficient for anything but text files though, storage wise.

[@4_real_bruh]

@@phizc true, but it is useable

[@AndersJackson]

@@phizc and lot of stuff is also just text files. This is why binary config files are the spawn of the devil

[@GreatLakesDrifta]

GIT is terrific. A feature my team uses extensively is Pull Requests, great for quick code reviews. Also, git integration with IntelliJ/IDEA IDE tools is great for finding potential breaking changes when "nothing works anymore".

[@anujkondhalkar9776]

git branch -M main ? What is it new in github. -M and -M main. Please tell me.

[@alice_in_wonderland42]

@@anujkondhalkar9776 why tf r u spamming this same qs everywhere?

[@anujkondhalkar9776]

@@alice_in_wonderland42 , Because I had doubt ? So I send this question everywhere, So I can Get some help. I am new in git. Git just updated last year 2021 thats why.

[@RubikMaster13]

Pull requests?? How does that work? We use merge requests for code reviews.

[@code123ns]

Pull requests are not a part of git

[@jenna_tools]

I never realized how critical Git is until I was had to use it professionally. Now I'd never go without. Git it the ultimate undo button. It gives the freedom to experiment and break things with reckless abandon, always knowing that you have the git repo to fall back on if you brick the whole app. 10/10 would recommend.

[@GytisValatka]

Man, I wish I had the same understanding of GIT so I would feel comfortable using it and exploring its potential.

[@menachemsalomon]

Would you consider having someone on to explain the basics of how "diff" works? Considering that "diff" is pretty much the ancestor of every SCM system in play (back from the SCCS days and CVS days), I would be really interested in understanding how it works.

[@oskrm]

diffinitely a good idea

[@gloverelaxis]

really it's `diff` that's the secret sauce for all VCSes. without diffing and its inverse operation (patching?), they'd all be utterly useless: no automatic merging (thus no easy branching) and impractically large history files.

[@dan-nutu]

Menachem Salomon: Three quick suggestions: 1. Wikipedia article. "The algorithm [...] became known as the Hunt-Szymanski algorithm". In a nutshell: "The operation of diff is based on solving the longest common subsequence problem" 2. man diff (the usual "read the fantastic manual") 3. source code Happy New Year all!

[@menachemsalomon]

@@dan-nutu I've tried reading the wiki article. But I don't have the necessary mathematics and computer science background to understand it. Hence my asking for a video. As for the fine manual (that's the way I learned it), I don't recall either the man or the info page explaining how the algorithm works. The source code might be an option, if I got desperate.

[@AndersJackson]

Actually, I recently discovered that git doesn't store any diff, like most other SCM systems. Git store all files that are current and in existence when you do a commit. Git never stores any diffs (conceptual). Files are compressed and then stored in git internals in full. And internals of git is really not that hard too understand.

Fun Fact: Linus Torvalds, the "inventor" of Linux also invented git.

[@Michael_Smith_1]

Me: I want a Mike Pound video Mom: we have Mike Pound videos at home. This: **Mike Pound video at home**

[@RonJohn63]

No mention of the phrase "version control", much less that it's *distributed* version control.

[@kwanarchive]

He does mention version control. And it's pointless for an overview to talk about the differences between distributed and non-distributed.

[@RonJohn63]

@@kwanarchive I thought I paid attention to the video, but don't remember him saying that.

[@kwanarchive]

@@RonJohn63 He literally says "version control" at 0:03 .

[@AndersJackson]

@@kwanarchive and that is also correct. Git is a version controll system with support for being distributed.

[@lawrencedoliveiro9104]

9:20 Unlike Subversion, Git does not keep track of what you have checked out at the commit level, only at the branch level. To get around this, it can be helpful to use the “checkout -b” command to create a new, temporary branch pointing to the commit you just checked out. Since this branch is considered “fully merged” (the commit is part of the past history of some branch you really want to keep), you can delete it afterwards with “branch -d” (lowercase d, not uppercase D).

[@joegee2815]

We used to use SCCS back in the 1980s on Unix.

[@AndersJackson]

Oh, I remember using SCCS too.

[@kosnk]

I wonder if governments use some version control systems when they update the laws.

[@calinguga]

i can tell you for sure some countries do not.

[@oresteszoupanos]

The parallels between git timelines and films such as Back To The Future, Looper, Time Bandits, Primer etc. could be used to teach git to each film's fans :-) but it may be too niche as a teaching tool for Computerphile's broad audience...

[@michaelcumming1233]

I've often thought that too: wouldn't Git be useful for tasks and concepts way outside of software development?

[@JosephDavies]

@@michaelcumming1233 Absolutely.

[@AndersJackson]

Might you give us an synopsis of the manuscript of that?

[@FFVison]

I remember using SVN many years ago for a web development firm, though I don't know all the details of how that worked for them, but was able to work it enough to do what I needed to do. I think they switched to git shortly before I left.

[@AndersJackson]

Before git, svn was the best free software tool to find. Then came git. (The same way rcs replaced sccs, cvs replaced rcs, and svn replaced cvs)

[@MrVipulLal]

Really basic stuff. There are thousands of videos at this level. Wish you would have gone deeper

[@calinguga]

casually propping a stringed instrument just slightly out of frame, such a classic

[@BrianWoodruff-Jr]

On the topic of rebase vs merge, I'm in camp rebase. I keep my feature branch up to date with the develop or master branch, taking care of any conflicts as needed. This keeps the master branch linear. There are no merge conflicts because those are taken care of during rebasing and the developer who wrote the code can take time making sure everything works before submitting a merge/pull request.

[@lawrencedoliveiro9104]

Rebase is OK for private branches. Once you publish a branch, rebase is going to cause those who pull that branch a lot of grief.

[@pianochannel100]

This wasn't a very good explanation of git. Felt pretty disorganized, and for that reason failed to inform.

[@bshulke]

Edit: NVM. It's an iOS feature to change the date to the day the iphone was originally announced when screen sharing. The iPad has a date of January 9th. Either they've been sitting in this video for almost a year before uploading, or are they're from the future.

[@BAgodmode]

That thumbnail is perfect. Picture of a guy with GIT right next to his head.

[@dcc1165]

This is a VERY misleading video. You're implying files are "moved" from the main area, to the staging area, to the "git" area (via commit). Git doesn't do this at all...it only tracks file changes (diffs) of what's being managed. The only time Git actually copies code to the "git" or "staging" areas is on an initial add/commit. All other activity on these tracked files is stored as diffs between commits. Also, code protection is only a minor side-effect/benefit of Git. It's only a safeguard if you commit very very often. If you make a ton of changes between commits, rollbacks are horrendous because it may revert you "too far back". Git was originally designed by Linus Torvalds as an alternative to SVN/SCCS/RCS/etc to allow for "detached" collaboration of code. SVN/SCCS/RCS, etc. do not allow for simultaneous code changes. You have to "check out" whatever you're working on, which locks the code to read-only for other users. With Git, multiple people can be working on the same code file(s) by cloning the repo. Each of their changes will be merged without hassle...unless the same line of code is changed by two different people, at which time Git will alert you of a merge conflict. At that point, you have to contact the "other" author and agree on whose modifications are the best implementation.

[@rezmed1144]

The best GIT explanation

[@lawrencedoliveiro9104]

By the way, fun tip about the “du” command: if you type something like du -ks . that shows you how much disk space is being used by the current directory and its contents. Suppose it contains a .git subdirectory,

[@lawrencedoliveiro9104]

and you want to see how much space is being used just by the current source code, not including the version history. If you do du -ks .git . then it shows you the space used by the .git subdirectory, and then it _subtracts_ that from the usage it shows for the “.” directory. Because it knows the earlier directory is contained within the later one!

[@jozefws]

best host of computerphile

[@ReviveNRepair]

Great video! I think it would have been a bit easier to follow along if the example was demoed in the terminal instead of drawing :) And version control software should definitely be taught alongside every introductory course in programming.

[@TJoseph2]

Interested in that electric double bass in the background

[@animarain]

Give us more Dr Max!

[@IVAN_ENT]

Git add

[@Danielagostinho21]

What a coincidence. I just started using Git yesterday

[@thargy]

Honestly, this was not a great explanation and there were quite a few (being generous) misleading (at best) statements. Not one of the better computerphile videos sadly, and I wouldn’t even recommend this as a basic introduction to git.

[@thargy]

I really don’t want my comment to come across as too critical- however, YT does insist the comment section is the alternative to the downvote. There are genuinely misleading comments in this video and it should be avoided by anyone trying to understand git. For example, implying SVN (aka Subversion) and Git are fundamentally the same, is true at a very insubstantial level - but it would have been better to acknowledge that Git is a Distributed Version Control System, whereas SVN is fundamentally a VCS - and the distributed nature.of Git is fundamental. The video is riddled with these. It doesn’t really explain staging well. Labelling a branch based on the point of branching (e.g. “when it was working) as opposed to what is for (e.g. “fixing issue 18”) demonstrates a lack of real world experience. There are so many poorly-phrased claims that I really do think it should’ve been better proofed. Again, I really respect Computerphile, and this is the first video that I’ve felt compelled to warn others off.

[@ryangrogan783]

My only gripes with git is that so many benign error crop up. I've had entire projects break or someone's branch for some reason doesn't ignore a file. Really frustrating when you can't push or pull because of some internal error somewhere that requires in depth console commands to fix. Not ideal when your team doesn't really understand the nuances or commands.

[@c-LAW]

0:14 what is git for? It seems to be a great place for posting software license keys and many other hidden truths.

[@123TeeMee]

So yeah, you need to stop thinking of "files", and more of "file changes". A bit like x versus dx in maths. A mistake I've made is changing and then staging a file, but then changing it again (before committing) and forgetting to stage it again, so I got strange errors due to the missing changes. In terms of software development in industry, git is essential. You should probably use a tool that gives you a graphical interface for git, like GitLens for VSCode, or GitKraken. Having the most common commands as buttons and being able to glance at a list of your changed files, whether they're staged or not, is very useful.

[@AndersJackson]

Or use magit, which is great.

[@bgezal]

I used CVS 20 years ago and the basics were the same. I think this wasn't as much of Git introduction as a versioning system introduction.

[@lawrencedoliveiro9104]

CVS tracked changes on a per-file basis. So if you refactored your code and moved a function from one file to another, that had to be recorded as two separate commits. The problem with this is that, in principle, you have a state of your repo in-between those two commits, where the code is in an inconsistent state. SVN improved on this by allowing the recording of changes to multiple files as a single commit.

[@bgezal]

@@lawrencedoliveiro9104 I remember it being a pita to check in code and merge.

[@peterisawesomeplease]

@@lawrencedoliveiro9104 The two key advances over SVN are that its distributed and its branch based. Although 90% of people don't use its distribution abilities. And honestly 50% of companies don't even meaningfully use its branch capabilities.

[@parrotraiser6541]

Programmers work with directories. "Folders" are for desktop users, spreadsheet jockeys, &c.

[@barneylaurance1865]

It's the same thing.

[@parrotraiser6541]

@@barneylaurance1865 Of course, which is why there's no need for a misleading metaphor,

[@joshwilliams7692]

Something I’ve found really helpful is permanently storing the undo history in vim. I often need to revert to previous versions of my file that I didn’t commit to git. If you enable the option, vim will track every single change you’ve ever made to your files. It’s incredibly helpful.

[@V3racious3]

Why show when you can tell... Excellent use of the medium.

[@josephmcandrews9050]

Doctor Wilson, I really like your shirt! Thank you for another great video.

[@Lion_McLionhead]

Lions have been learning the quirks of git ever since it came out. Every startup has a different dance which exposes a different set of quirks, a new set of goofy finnish translations for words that svn, perforce & cvs used.

[@michalbotor]

i believe that it is important to emphasize that git does not store **copies** of your files in different states - that would be wasteful and not much better than you doing it manually - what git stores are **differences** of your files in different states, and so what you see when you check out a commit is not the state of the commit but the stack of all commits up to that point.

[@godarklight]

It might have been a bit clearer to actually show the git commands and the folder view rather than use the drawing pad

[@emmanuelasika1590]

About to go full-stack so this video came at the right time.

[@morosis82]

Git is useful whether you're full stack or not.

[@emmanuelasika1590]

@@morosis82 yeah... Just, I've mostly been doing WordPress related gigs. Now looking to go full-stack from polishing my frontend skills and then learning new backend tech. Git is a stable in this case

[@sandyjonesmusic]

Nice overview, thanks dudes. BASS!

[@ThatBeTheQuestion]

I really wish someone had taught me this when I was in school....

[@TesterAnimal1]

Git makes life *possible* for distributed teams.

[@keescouprie5968]

Why does nobody ask about the instrument in the background?

[@amarissimus29]

Even my computer fell asleep on this one. Cripes.

[@stephane184]

Reading up on how blockchain works kinda reminded me how Git works as well with commit hashes and version history.

[@AndersJackson]

@@__Brandon__ blockchains doesn't need to have prof of work, there are other ways of decide where the current top of the chain is in a secure way. Like a group if servers that you can ask and the majority will point ro the right node.

[@knightcrusader]

Git is a great tool, and the way it structures data is brilliant. However, the command line interface is a goddamn mess. The commands are not intuitive at all, and I feel this is the biggest hurdle for getting people to understand it.

[@user-hw3fz7ic5l]

نعم

[@mptystring7892]

I agree, command line interfaces scare people. It's sad when people stray towards graphical Git clients. For some, the GUI "training wheels" might never come off, limiting the ability to collaborate and use Git efficiently. Thankfully, work is being done to make Git's CLI more friendly. Some commands have been renamed so that they better describe their actions, instead of being named after similar commands in other version control systems. For an example, `git switch` is a new alias for `git checkout`. The command has always switched the working tree (the files in your git folder) to that of another commit. It has never "checked out" files, since there is no central repository server managing who has which files checked out (=locked for only them to edit).

[@AndersJackson]

@@mptystring7892 never seen it as locking when checked out. Switch work too. My myself, I uses magit in emacs. Really powerfull git management tool.

[@Nalojuntu]

Ok well... I'd really like to also see a video from you about how Git works from the inside, I mean the internals of it. : )

[@whoareyouandwhatdoyouwant]

Where can I buy the "three part slogan" t-shirt? Made me laugh out loud haha

[@Computerphile]

Search "Larry & Paul " ; ) -Sean

[@chingfool-no-matter]

Thanks for the nice explainer video! would be great if we could also cover the conflict resolution algorithm in the next one.

[@MichaelKingsfordGray]

How does GIT resolve: two simultaneous changes, from two different time-zones?

[@barneylaurance1865]

Git doesn't care much about time or time zones. It's interested in the order of things as recorded by one being based on another, not what time they happened. If two things happen in parallel it doesn't matter which one was first. Git will try and merge the changes together, if they conflict it will ask the person operating it to make up the synthesis of the versions that they want.

[@ericmarcelino4381]

I'm proud to say that I understand everything in this video!

[@Wawet76]

The story behind git creation is one that desearve to be told on this channel.

[@apid4075]

Git is simple and easy to use. If I didn't already, I wouldn't know what git is after watching this confusing video. I recommend looking up a better explanation.

[@discretelycontinuous2059]

This channel isn't really educational as such, rather entertainment / somewhat-informative

[@insu_na]

Don't forget the glorious `git rebase` all hail the `git rebase`

[@NnO0Worries]

What's up with the giant music instrument in the corner?

[@markostojanovic6973]

great topic, such a pity this was a pretty bad presentation (if i didnt know what he was talking about i prob would only get scared of using git). but on the other hand its impossible to explain git in 15 mins

[@JeanNoelAvila]

Quite frankly, moving file pictures around does not give an accurate explanation, and is even misleading.

[@peterisawesomeplease]

Yea it is super difficult to describe and 50% of organizations honestly don't use any aspect of git that svn was not capable of.

[@hashomi0596]

Great video. Git for me is a tool to go crazy and break stuff then revert.

[@RandomNullpointer]

Can I use git with Libre Office documents?

[@thadtheman3751]

SAdly, his log looks like the logs of of VC atcompinies I worked at.

[@aryanilpanja9846]

he really explained git well!!

[@code123ns]

Once you get proficient with git, once you really understand how the internal data model works and how the commands manipulate it, you will be able to solve almost half the problems that git creates, without any trouble. Also, never forget that git has two kinds of commands: plumbing and porcelain and that neither is targeted for end users.

[@AndersJackson]

Well, the only problem is to locate which command and sub command with what arguments that you should use. 🙂

[@noswonky]

I've been using git for a few years amd before that Mecurial, SVN, CVS, RCS, ClearCase, Perforce, and SCCS, but this video just confused me.

[@sageinit]

Next, do pijul, and how it (and it's spiritual PREdecessor, dArcs) differs tremendously from git

[@AndersJackson]

Why? The presentation and explanation of git is in no way even finished on basic level.

[@oldcowbb]

git gud

[@willemvergouwen]

at git

[@drdca8263]

'gud' is not a git command. See 'git --help'

[@hectortorres6567]

This was really informative thank you

[@thecaveofthedead]

Speaking of collaborations, when is a musical one going to happen between your presenters?

[@felicianomazoio8012]

I want to be smart like you guys! 🥺

[@refactorear]

2:28 Yep, but thanks to TDD now I know exactly when it breaks :)

[@anttikalpio4577]

I came here thinking you’d explain differencies between subversion, git and if there are any other comparable options. Please! Why git and not svn?

[@AndersJackson]

For one, svn need internet access to work against the repository. In git your machine has your repository, so you have all history with you. So you can work on an airplain without internet access. Git can work in many different way, when it comes where repositories are stored. So it is much more flexible. You can for instance sync against many different repositories. Svn have problems manage directories, which git doesn't have. So yes, svn is basically only usable for history reasons. Anything svn can do, git can do. And probably better. That is not the case in the other direction. Yes, I have worked with RCS, CVS, svn and git. And that is my experiance. And each software is superior to the previous.

[@abhirishi6200]

Cool video!

[@veenified]

Dive deeper into the technical details oh how git works.

[@PushyPawn]

Oi! Who you calling a Git mate? Blimey!

[@petterjunni]

Fun fact: both git and the Oura ring that Max is wearing are made by Finns.

[@barneylaurance1865]

Git was started by Finn, but the current maintainer is Japanese. And of course hundreds or thousands of people from many countries have contributed to it.

[@marcogobbi0]

great overview of how git works. thanks a lot. Can git be used, somehow, to create a collaborative application? i mean... imagine a web app where multiple users can edit the same data at the same time. just a crazy idea that come out of my mind :D... git proved to be a great tool for developers team work, so why not applying the same concept to an application that implies business guys working on the same data

[@Daviid_5]

Never thought about it, can you have git without a remote repository? As in 'git init' and then only using 'git add', 'git commit' and the like? Without git push? For some reason I've always created a "remote" repository in a different folder to "backup" my projects (same hdd, really)

[@glasket_]

Yeah, that's viable. You don't need a remote; all repos are fundamentally the same, so your instance of having your remote in a different directory on the same machine is effectively just saving the same info twice for no gain. The only real situation that you would *need* a remote for is having a canonical source to share code from (technically possible on your own machine, practically a *horrible* idea though).

[@barneylaurance1865]

Absolutely. If you just need a backup that doesn't have to be a remote git repository. You can just use a regular file backup system.

[@PauloConstantino167]

lol presenting git commands orally. i guess I should emulate a cpu in my brain to learn git then

[@dankrukov1890]

engineering masterpiece one yet to find anything nearly as unbelievably perfect (may be linux kernel)

[@AndersJackson]

Unification algorithm and resolution algorithm behind the Prolog language might apply for that.

[@bassett_green]

I want to know about that hotel desk bell on the sill behind him

[@rylaczero3740]

Also make a video on Mercurial and how its being adopted at bigger companies.

[@lawrencedoliveiro9104]

If they were using Microsoft Team Foundation Server, you can understand why _anything_ would be seen as an improvement on that ...

[@davidadams421]

Need more.