Linus on GIT and SCM

An anonymous reader sends us to a blog posting (with the YouTube video embedded) about Linus Torvalds' talk at Google a few weeks back. Linus talked about developing GIT, the source control system used by the Linux kernel developers, and exhibited his characteristic strong opinions on subjects around SCM, by which he means "Source Code Management." SCM is a subject that coders are either passionate about or bored by. Linus appears to be in the former camp. Here is his take on Subversion: "Subversion has been the most pointless project ever started... Subversion used to say, 'CVS done right.' With that slogan there is nowhere you can go. There is no way to do CVS right."

Source Safe

[EraserMouseMan (847479)]

Well Linus didn't have anything bad to say about MS Source Safe. . .

[ducking] Sorry, I couldn't resist the urge. ;-)

Why winge?

[gilesjuk (604902)]

CVS and Subversion are open source projects, Linus should fix them.

Re:Why winge?

[zzatz (965857)]

Linus isn't saying that CVS and Subversion have fixable bugs or missing features. It's not about the code.

He is saying that they solve the wrong problem. The Subversion team wants to solve Problem A, and Linus wants to solve Problem B. No amount of code will turn the solution to Problem A into a solution for Problem B. Bothering the Subversion team with code addressing Problem B will only irritate them, since they're working on Problem A.

The right way to handle differing goals is to start a different project. That's what he did.

Don't be confused by the labels. Source Code Management means different things to different people, and there isn't always much overlap in how each person defines it. Ships and airplanes are both 'vehicles', but that doesn't mean that a few changes will turn one to the other.

Re:Why winge?

[RedWizzard (192002)]

CVS and Subversion are open source projects, Linus should fix them.

He did fix them: he wrote GIT. He's no really whinging, he's saying "I wrote this tool because the other options are crap".

how to learn git?

[zojas (530814)]

I've tried to use git, and I feel like if you want to do anything more than commit, you have to jump off a cliff which has serious spikes at the bottom. seriously, if you want to learn how to do more than 1 or 2 of the simplest operations with it, you have to invest serious time. I tried, and never could get there.

anybody have a good tutorial? (not the crappy one which comes with it)

I'm not an SCM rube either. I've competently used tla (arch), darcs, and of course CVS. but git just seems too hard to use. damn fast though.

Re:how to learn git? - answer, don't!

[Omnifarious (11933)]

My favorite, of course, is Mercurial [selenic.com]. My main draw is that I had been interested in distributed SCMs for years, but had never found one that made any sense to me whatsoever. I was on the hunt again and stumbled on Mercurial, and I've been hooked ever since.

Of the various distributed SCMs, Mercurial is the easiest to use one I've found. And it's pretty fast, though not quite as fast as git (though I have some ideas on how to fix that). And since it's written in Python with only a very small C component it runs on many platforms.

Re:how to learn git?

[by Anonymous Coward]

# set up new project
cd project
git init
git add .
git commit -a -m "Initial commit"
 
# edit a file
vi file.php
git commit -a
 
# add a file
vi new.php
git add new.php
git commit -a
 
# see the log
git log
 
# make a branch
git branch working
git checkout working
# or in one step
git checkout -b working
 
# add some changes to this branch
vi file.php
 
# see what you changed
git status
 
# check it in
git commit -a
 
# see all branches
git branch
 
# go back to the first branch (initial branch is called "master" by default)
git checkout master
 
# make some other changes
vi other.php
git commit -a
 
# merge the working branch into this one
git merge working
 
# see the branches and merges in a graphical browser
gitk --all
 
# let's do a log of all commits in "working" that don't exist in "master"
git log master..working
 
# hmm let's undo that last merge (tip of branch is HEAD, one commit back is HEAD^.. we are "popping" one commit)
git reset --hard HEAD^
 
# push your changes out (push the tip of local "master" branch to remote "incoming" branch)
git push foo.bar.com:~/myrepo master:incoming
 
# pull changes from another repo (remote "feature1" into local "feature1" branch)
git pull baz.bar.com:~/otherrepo feature1
 
# move the branch point of the "working" branch to the top of the "master" branch
git checkout working
git rebase master

It can get a LOT more complex of course.

When you're starting out, just remember "git commit -a" and you'll be fine. Also check out "git reflog" to see the linear history of your repo. The pulling/pushing stuff can get a lot more complex but it's damn powerful. If you can figure out Arch (yeesh) you can figure out git!

SLASHDOT SEZ: you have too few characters per line. Okay, slashdot, here's part of the man page for git-rebase:

If is specified, git-rebase will perform an automatic git checkout before doing anything else. Otherwise it remains on the current branch. All changes made by commits in the current branch but that are not in are saved to a temporary area. This is the same set of commits that would be shown by git log ..HEAD. The current branch is reset to , or if the --onto option was supplied. This has the exact same effect as git reset --hard (or ).If is specified, git-rebase will perform an automatic git checkout before doing anything else. Otherwise it remains on the current branch. All changes made by commits in the current branch but that are not in are saved to a temporary area. This is the same set of commits that would be shown by git log ..HEAD. The current branch is reset to , or if the --onto option was supplied. This has the exact same effect as git reset --hard (or ).If is specified, git-rebase will perform an automatic git checkout before doing anything else. Otherwise it remains on the current branch. All changes made by commits in the current branch but that are not in are saved to a temporary area. This is the same set of commits that would be shown by git log ..HEAD. The current branch is reset to , or if the --onto option was supplied. This has the exact same effect as git reset --hard (or ).

Well, speaking from my own experience...

[RootsLINUX (854452)]

I've used CVS, SVN, and GIT in serious projects and I can say I far prefer SVN to GIT, and GIT to CVS. GIT was incredibly confusing to use, and it may just have been the way the repository was administered was poor, but I never knew if I was synched with everyone else's checkouts and the command names made no sense. Its been over a year so I don't remember the details of GIT, but I remember having to do a lot of things "twice". Need to do a checkout? Two commands. Need to commit? Two commands. It was a bitch to use and I am glad I'm done with it. SVN, on the other hand, I felt very comfortable with from the start and most important of all, I trusted SVN to do what I wanted it to and to keep me from screwing up. In a year of using it, it has failed to lose my trust.

I'm not trying to say SVN is better than GIT. The best repository depends on the type of project and type of development. But defaming SVN in favor of GIT is not, I believe, a valid statement. Especially when (I'm pretty certain) many, many more projects use SVN rather than choosing to use GIT.

Re:Well, speaking from my own experience...

[Black Acid (219707)]

Its been over a year so I don't remember the details of GIT, but I remember having to do a lot of things "twice". Need to do a checkout? Two commands. Need to commit? Two commands. It was a bitch to use and I am glad I'm done with it. SVN, on the other hand, I felt very comfortable with from the start

Most distributed version control systems exhibit this phenomena, because by "checking out" you are actually doing two operations: pulling the latest changes from someone else, and updating your workspace. For example, in Monotone you would type (I imagine git operates similarly):

mtn pull
mtn update

The first command retrieves revisions from the server, and the second updates your workspace with those new changes. To "commit" a change, in a distributed version control system you first 1) commit the change to your local repository and then 2) push it to someone else:

mtn commit
mtn push

It is often useful to keep these operations separate. For example, you can commit without pushing. Make a bunch of changes, commit each one separately, and only push once you're satisfied with the result. Other developers can still see each change you made individually, but only after you've pushed, so they won't be stuck with an incomplete in-progress version of the tree.

Similarly, by being able to update without pulling, you can revert to any revision you would like without contacting the network. Likewise, since commit does not require network access, it is no extra effort to work offline. Once an Internet connection is available, you can synchronize your repositories, but in the meantime you can make any change you want - even with no network connection.

The main disadvantage of a decentralized version control system is that it requires workflow changes [pidgin.im] to get the most out of it. If you are only familiar with centralized version control systems, it will take some time getting used to. But I'm glad to say, an increasing number of projects are making the change to distributed version control [slashdot.org], among them, Mozilla and Pidgin. They are not using Git (but Mercurial and Monotone, respectively) but they're all distributed. Git is being used by the Beryl [beryl-project.org] project, among others. Subversion has momentum in FOSS because it is familiar for those used to centralized version control (everyone knows CVS), and SourceForge [sourceforge.net] provides free SVN hosting. Once a free open source hosting site provides hosting for a distributed version control system, I expect more low-resource open source projects to use it.

There's a difference between GIT and SVN

[paroneayea (642895)]

... And that is that CVS/SVN are centralized, while GIT is distributed, like GNU Arch.

There are appropriate uses to both of these, and in kernel development I think it makes sense to have distributed development. However, in smaller projects, which really *need* a very specific direction (example, Wesnoth, I would think would not have gotten where it is today if there were so many branches where people were all making their own art).

Linus is enough of a famed leader that he's going to be listened to, and thus kind of pulls the community around him as a central source of development. That's not necessarily going to happen everywhere.

git is pretty cool, take a closer look

[by Anonymous Coward]

I took a look at git a while ago and was completely underwhelmed. The UI was so bad it was useless, and it didn't "seem" to do anything that Darcs didn't do. (I used to love Darcs because of the automatic patch dependency computations).

Now that all the "next generation" SCM tools have matured somewhat, I took a look at all of them again. I had to stop using Darcs because of the "patch of death" problem, which basically is this: after using Darcs on a project with long-lived parallel branches, the repository may eventually enter a wedged state you can't get out of, due to exponentially complex patch dependencies. Oops.

At this point I had an idea of what an SCM should do, how it should work, what the "mental model" should be. I want to create changesets, add them to branches, combine multiple branches (and keep track of renames and so forth between branches), re-order changesets, collapse multiple changesets into one, discard old branches, etc.

Of course, CVS and close cousin Subversion are SO UTTERLY USELESS I didn't even consider them. Seriously, Subversion is like gold-plated shit. Looks nice but it's still shit. Reading people say stuff like "Subversion is awesome" makes me wince. How can something that doesn't have "real" branches, and doesn't have tags OF ANY KIND, be useful for anything? How do you keep track of multiple merges between branches? Answer: you don't. Or you keep track of revision numbers using svnmerge and pray it all works. Even the Subversion docs sortof hand-wave this away. I.e., they hand-wave away one of the FUNDAMENTAL ASPECTS of source code management: branching and merging. It's like hearing people talk about OO databases. They mean well but they just don't comprehend the generality of the underlying problem.

That's why I was so excited about Darcs: the author "gets it". Unfortunately the implementation is flawed.

I checked out a few more (Mercurial, bzr) but finally settled on git because it let me do all the things I needed to do, and it did them FAST. Once I figured out the underlying model I was pretty impressed. Git can be viewed at many levels: very low-level plumbing, or UI-level, or in between. The UI and documentation is still pretty shitty, but thankfully they are working on improving it and are moving away from the idea of having interchangeable UIs. Just focus on improving "core git".

One great thing about git is that so much of it is just files in the .git dir and shell scripts that combine very simple low-level functions. For instance, you can create a branch just by saving the SHA1 ID of the tip into a file in .git. You can branch off any point in the history this way, including branches you've deleted in the past (git keeps all the old commit objects by default, even ones that aren't pointed to by any branch or tag.. this is very simple and understandable model, like reference-counting in a way).

The other great thing about git is how easy it is to sling changes around and reorder them and combine them. For instance let's say you add a file to your project as commit "A". Then you add some code that uses this file as commit "B". Then you fix a bug in the file as commit "C". So you have A-B-C. Now you'd like to combine A and C into a single patch A', and put B on top of it, like this: A'-B. In git, this is super-easy. I can think of two ways to do it off the top of my head.

I was checking into a CVS project the other day (for a client) and wanted to do this. Then I realized, you can't move things around in CVS like this *twitch*. So nowdays I do everything in git and only after the changes are beautiful and self-contained and well-commented do I check them into CVS one at a time.

Okay so they point is, check out git (or honestly? Checkout out ANYTHING that isn't CVS or svn). Even if you think Linus is an asshole (which he is) or you don't like the git UI (it's not that bad now), check it out anyway.

And if you don't use SCM at all? You suck. Start learning. It's a best practice that you can't live without, once you start.

Re:git is pretty cool, take a closer look

[Senjutsu (614542)]

Making a copy is not the same as making a branch. ... And for fucks sake Subversion, creating a copy in a directory called "tags" is not the same as making an actual tag.

The way subversion does "copies" (there is no duplication of shared data between copied directories), there is no difference in practice.

Distributed version control gaining ground in FOSS

[Black Acid (219707)]

The ultimate reason why Linus dislikes SVN, CVS, etc. is that it is centralized. Everyone checks out source from a central server and commits their changes to the same centralized area. This has problems: your workspace is not versioned. By this I mean, you cannot track local changes to your workspace without committing them to the central server.

A common pattern in development is to try one approach, test it, tweak it, and possibly try another approach if the first did not work out, perhaps reverting to a prior approach. With decentralized version control, you can commit your changes to a local repository and work from there. All the locally changes you make are versioned, and be committed, checked out, examined all without contacting a central repository. This is ideal, because you often want to try various options to find the one that works best, before pushing your changes to the rest of the world. In centralized version control, you can use a branch for this purpose, but often branches in these systems are difficult to either create, merge, or maintain, so they are rarely used. The end result is that with centralized version control, developers version their workspace in their head. DVCS systems remove the mental burden.

Fortunately, FOSS developers are realizing the usefulness of DVCS and major projects are converting to some form of DVCS. Mozilla is switching to Mercurial [mozillazine.org]. The Pidgin [pidgin.im] project, which just released 2.0.1, is using Monotone [pidgin.im]. (Linus favorably mentioned both of these distributed version control systems in his Git talk, as they are both are distributed).

Once you accept that DVCS is better than the centralized model (which may not be true for some situations), only a few (but growing number of) version control systems are viable. This is currently a hot area in open source development, with software such as GNU Arch, Monotone, Mercurial, Git, Darcs, Bazaar, and more paving the way. Many open source DVCS's are still in development and not ready for general usage. I can't speak for Mercurial, but Monotone doesn't have the greatest performance, instead preferring integrity over speed. This led Linus to write git, since speed is very crucial for a large project like the Linux kernel.

Whatever the actual program (git, Mercuial, or Monotone), more and more open source developers are realizing the advantages that distributed version control can offer. I encourage all developers that haven't used any DVCS to try it -- once you do, you won't go back.

~$ mv CommitAccess MergePrivileges

[Excelcia (906188)]

Linus talks about his distributed model, how everyone has a branch, and how this avoids politics associated with who gets commit access. He claims (and I admit I've seen this happen in some) that many projects have quite the internal politicking on who has CVS commit access. But then he claims that Git's special sauce eliminates these internal politics. Ok, I was intrigued, so I listened on.

Essentially, he explains, the secret with Git is that everyone has commit access on their own branch - they do whatever they want. He says that the way it works is that someone does something cool with their own branch, then they start hollering to say "Hey, I have a good branch, merge mine" and it will get merged. Politics over.

Ok, so now I'm scratching my head. How is this a fundamentally different paradigm? In CVS, basically anyone can check out the whole tree and make any changes the like. They can then say, see, my changes are good and ask for them to get committed or ask for commit access themself. In Git, this commit access bottleneck is just moved from the commit stage to the merge stage. You make your changes, commit them to your separate and unique branch, and then ask someone with to merge it, or give you the ability to merge it in to mainstream. How exactly does this eliminate the politics? You are still going to have some people with "the power" and some people without. In any project where you have people who are going to fight about who gets commit access, you'll just have a fight about who has the ability to merge into mainstream.

So, ok, distributed is nice (though for some projects central may be preferred) but I don't see how this magic system bypasses politics. In fact, I can potentially see more internal politics over this method. I can see factions gathering to support this or that branch, arguing about which is better, fighting about which one gets merged in. I can see the potential for branches going longer between merges, and more changes happening at once, making it harder to track problems. I don't claim these scenarios are more likely, but I do claim that this changing from a commit access to a merge access paradigm is just renaming the problem.

Re:~$ mv CommitAccess MergePrivileges

[iabervon (1971)]

The advantage is that MergePrivileges can be fine-grained: there can be many answers for "merge into what?" There's a -mm tree, a -stable tree, a -linus tree, a -rt tree, and a lot of vendor and distro trees. Each of these has a different maintainer, and can have a different idea of what is acceptable. And only the maintainer can merge things into their tree, and they can decide based on a variety of features of the things they're considering. For example, Linus only merges from a few people directly: maintainers of various subsystems. And he doesn't even trust them completely; if the SD/MMC maintainer has a change which changes x86 architecture code in the tree Linus is asked to merge, he'll notice and ask what's up with that. And if there are changes that look too intrusive for the current point in the development cycle, he'll put it off until the next cycle, and ask for a tree with just fixes. And -linus isn't special, except that almost everybody trusts him implicitly and merges his stuff into their trees (the main exception being -stable, which is why a new 2.6.20.x kernel isn't derived from 2.6.21; and vendor and distro kernels are generally based on -stable of some sort, and only get new stuff from Linus when they go to a new series). Also, maintainers of subsystems know the people who work in their areas, and can apply the same sorts of rules: the guy from Intel who works on their network drivers can get e100 changes into the the -netdev tree, because the maintainer knows they know what they're doing for e100 changes. And Linus sees that the e100 changes are coming in through -netdev, and the network maintainer knows what policy to apply to the drivers around there, so they're fine, even if Linus has no clue who should be allowed to do what in e100.

It's not that the politics go away. It's that the policy is no longer a binary "yes or no" decision, so the technical arrangement mirrors the social arrangement. This doesn't work with CommitAccess because people wouldn't commit the same change everywhere they should, and they couldn't be restricted to only making changes they're trusted to make (there are people who are trusted to correct spelling in comments in any file in the tree, and Linus can look through the total changes they send and verify that they only change spelling in comments).

It depends on the project

[ClosedSource (238333)]

If you have a project that has thousands of developers all of the world like Linux does, a SCM system that is focused on merging makes a lot of sense. Unfortunately, there is a tendency for some people to overdo merging on small projects when they don't really need to. If the application is designed in a modular fashion and developers are assigned specific modules, than merging is rarely needed. Of course, many control freaks don't like this approach because it makes it harder for them to "correct" other developer's code.

it'll never catch on

[sohp (22984)]

Distributed version control the way git does it (conceptually, not necessarily the implementation) is the best idea in SCM since concurrent development and optimistic merge conflict resolution on check-in.

Notice how, even years after better ideas superceded the lock-modify-unlock paradigm, many tools and shops still use exclusive-lock SCM.

It could be quite a while before you see anything like the way git does SCM in use in the majority of programming shops.

Re:git

[by Anonymous Coward]

He is only human. Just because he is the head of a huge software project doesn't make him infallible.

Just look at the whole 'RMS vs Linus' thing.

His opinions should carry some weight, especially since he should know more than anyone what the limitations of SCM software is when it comes to larger projects like the linux kernel. But a lot of SCM comes down to the way a project is managed, the preferences of the people involved, and how they deal with their project. I doubt there is a blanket solution... a 'one SCM package to rule them all' so to speak.

Especially in the software industry you can always find someone just as good as yourself that strongly holds opinions that are the polar opposite of yours.

Re:Merging *does* suck

[Frankie70 (803801)]

So don't do it


Wow! I bet you have never worked on anything other than hobby
projects.

Most projects I have worked on cannot do without branching &
branching big & I am not talking about branches created for
individual devs.

What do you do if you have make patches on an earlier release(s)?
What do you do if your project team has 50 devs working on
5 different modules inside? If one guy makes a buggy submit
it will break every one else? Typically each team does weekly
sanity tests & then propagates the changes to the main.

Yeah - and I agree with Linus - CVS is rubbish.

Have used CVS, Clearcase & Source Depot. Source Depot
is a Microsoft internal Source Control system. Microsoft
licensed Perforce & developed on it. I used to work with
MS long back & Source Depot was the best Source Control
System I have ever used.

CVS lacks too many features.
1) Atomic checkins/submits
    I am trying to submit changes in 5 files as a single bugfix.
A submit/checkin should either succeed for all 5 or fail for all 5.
CVS doesn't do this. The end result is that I may end up submitting
a change in the header without submitting a correspond change in the
implementation file.

2) Changelists
    After checking in multiples files together, at any point in time, I should
be able to find out all the changes that were checked in at the same time.
CVS has no way of doing this - Submitting 5 files together is the same as
submitting 5 files separately as far as CVS is concerned.

3) More Changelist features for non-submitted changes
Let us say I am working on 3 different bugfixes. Source Depot allows me
group together my changes in different changelists even before I
submit the changes. That is I can create changelist A B & C.
In changelist A - I have files a.c & a1.c changed, in changelist
B, I have b.c & b1.c changed & so on. So I decide I am done with
all the changes required in the subset A, I can submit it very easily
or undo all changes in changelist B.

4) Merges
Merges between branches are a breeze with Source Depot. With CVS it's
a pain. Source Depot stores a lot of information about merges which have
already happened which in invaluable. In CVS, merges between branches
are very little more than changes manually copied from one branch to
another.
I can do a lot of stuff which I can't do with CVS
- I can very trivally merge Bugfix 1111 (comprising of 5 files
checked into changelist XXXX) from a branch to another branch or
the main trunk.
- Because Source Depot stores information about merges, I can do periodic
single command merges very easily between a branch & the trunk - Source Depot
will not try to merge in changes which have already been merged the last
time I did a merge.

I could go on & on, but the point is that something Source Depot makes
a developers life so much more easier. I could work around all these
things in CVS (i.e. do it in multiple steps) but the ease is something
worth paying for I think. If Microsoft ever released Source Depot
as a commercial product, it would be great, but I don't suppose their
license with Perforce would allow it.

Re:Merging *does* suck

[Timothy Brownawell (627747)]

So don't do it.

The Wise adapts himself to the world. The Fool adapts the world to himself [revctrl.org]. Therefore, all progress depends on the Fool.

It's always done late in a development cycle, in the rush to get the project out the door.

Why? It doesn't have to be. At least if you use something that isn't horribly broken [red-bean.com].

So don't branch, and DON'T allow concurrent checkout of any code - FORCE the DEVELOPERS who need to work on the same code to COORDINATE their work EARLY in the development cycle. Of course they'll bitch.

Yes, they will. Because this is a monumentally stupid idea. Because the entire *purpose* of revision control systems (note: "CVS" stands for "Concurrent Versioning System") is to make it possible for developers to work on things at the same time. The idea is that you can get more benefit from the concurreny than you get difficulties from merging.

If your technical leadership has the spine to show prima donna twits who won't follow development rules the door. Of the entire company.

Rules like "merge early, merge often", perhaps? Fixes the problem, and *doesn't* cripple development horribly like your idea would.

Re:Lemme check my last home appraisal...

[starwed (735423)]

You missed the point of the thread; to discuss git, not to be one.

Re:Lemme check my last home appraisal...

[Black Acid (219707)]

You hit the nail on the head. Distributed version control often comes with superior merging, making the process less painful and encouraging it to occur frequently. Monotone employs a 3-way merge [wikipedia.org], Codeville has an innovative merging algorithm [zooko.com], and some may even support 5-way merging [nongnu.org] ("left's immediate ancestor, left, merged, right, right's immediate ancestor") in the future.

In my experience, nearly all merges occur automatically and cleanly. Only if two developers modified code in conflicting areas of the source code do you have to merge manually--and even then, only one person has to do it. It is much better to have merging operate automatically and transparently when possible, than to have to have two people manually coordinate each and every one of their changes beforehand.

Re:Linus knows it.

[CastrTroy (595695)]

You might want to check out TortoiseSVN if you're using svn on windows. It makes version control really easy, and you don't even have to touch the command line.

Re:Linus knows it.

[statusbar (314703)]

I use SVN on windows, mac os x, linux (ubuntu, debian, fedora) as well as netbsd. TortoiseSVN works great on windows especially for the point and click style users who need to use SCM. SvnX works great on Mac OS X. Altium PCB designer works great with the svn command line tools and shows graphical diffs of our circuit boards. But for some reason, Tortoise SVN and svn.exe are unable to access a GIT repositiory.

In addition, git works well for simple projects but not so well for projects that have many different related subprojects which share code.

  For instance, our SVN repository holds everything needed for an entire product, including embedded linux with busybox, initrd and custom software and libraries - as well as DSP source code for two different add on cards, the GUI for mac, windows, and linux, the docutils xml file for the various manuals, and manufacturing and test code.

I'd love to use git once it attains the required maturity level so that I can do what I need with it.

--jeffk++

Re:Well, Linus is an ass, what's new.

[True Vox (841523)]

Well, probably those two go together: being an amazing creator, and being an amazing ass with huge ego. Who knows.

I disagree entirely that those two traits must go together. I'm living proof that you're wrong, in fact. I don't have a creative bone in my body.