Or is it a solved problem and Git is the endgame of VCS
I’m not that old, but I’m old enough to have used RCS, CVS, SVN, then finally Git. I started using Git super early, before GitHub existed. You may not believe me, but Git was the answer to all my prayers. All those previous systems had fundamental architectural flaws that made them a complete nightmare to use on a team for any serious work.
Git has no such problem. There’s nothing it can’t do. Instead, the only limitation is on the user being able to know how to get Git to do what they want it to do. Thankfully, that’s always solvable with a quick read of the documentation or web search.
I understand that people might want to make it easier, since the UI is complex, but that’s never going to work. If you abstract away the complexity of the UI, you will necessarily abstract away the power. If you abstract away the power, you are no longer solving all the problems that Git solved. People just don’t realize what problems are being solved because they never lived through those problems with the previous VCSes.
You know what’s easier than building a new VCS better than Git? You know what’s easier than using a different VCS from the rest of the world? You know what’s easier than designing a miraculous abstraction on top of Git?
Learning Git.
Whatever effort you are putting into seeking or building an alternative, instead put that effort towards becoming a Git expert. It will be a lot less effort with a lot more benefit. Trust me on this. It will be well worth it for your career and for your personal computing life as well.
Working with an engine like Unreal Engine for a project of any reasonable size requires working with both hundreds of thousands of active files (my current project's repo's HEAD has ~400k) and hundreds of gigabytes of active files, many of which are many GB on their own. Our repository (in perforce) is currently in the order of 10TB.
Git, even with LFS and partial clone and shallow copies and fsnotify just falls apart at this scale. Add to that the necessity for less-technical people (artists, designers, VFX, audio, etc) to use it, and it is really just a non starter.
I absolutely loathe Perforce (having used and occasionally admin'd it professionally since 2007), but I begrudgingly admit that is is currently the only publicly available VCS that can fulfill all of the requirements of this industry in a practical way. Plastic and the others just aren't there yet.
Most of this was built by Microsoft to work on Windows with git.
We moved from SVN in the late 2010’s, and holy crap man, did it change our workflows. No longer were we passing around patch files in a team for code reviews, but actually reviewing code somewhat like on GitHub. It was magical.
It's great that Perforce works and I've heard others in the graphics field using it, so it satisfies a need. Don't know if/when it would be a general need for say GitHub users.
If you have no problems with Git, then I’m happy for you. I certainly have problems with Git, eg its inability to handle large repositories meaningfully (enjoy a Chromium checkout!), the arcane CLI commands, the shut-your-eyes-and-hope of moving commits between branches and other sharp edges.
That Git has been as resilient as it has is largely a function of being written by Torvalds and GitHub network effects. It isn’t for lack of better/as good features found in other VCS methodologies.
To take your example of moving commits between branches, the git rebase parameters and --onto flag are definitely confusing, but each of them is clear and necessary when understood. The CLI could be refactored to be clearer, but the hard part is understanding what each one means, which you would need to do anyway even with a better CLI. Obviously one can work around these knowledge gaps, but once you understand them they are useful for an entire career and the warts tend to fade in importance relative to the value of a tool with the insane power-to-weight and rock solid architectural tradeoffs of git.
Nope.
So slow. It does not like millions of files. And all my tooling that does a “quick” git status locked up the terminal, vs code, etc.
1. The client diff content of your commit with previous one. 2. Apply such diff on content of another commit. 3. Create a new commit with the old commit message/author of commit in step 1.
And that is also what make git work for everyone, because it is simply a directory that contains tons of new_new_updated_project on steroids. It's up to you to decide what to do with it.
Out of curiousity, what other SCM tools can pull this off better?
Are the syntax of the commands the issue or are all CLI commands arcane to you? What would you propose to be the command syntax if you were designing it?
I disagree, really.
There is not fundamental reason why you cannot have user-friendly UI for 98% of the cases and some "advanced" for those 2%.
Just like GUIs have "advanced" settings, the CLI can have well designed 10 commands for basic usage that are taught to newbies and then advanced ones.
And if anyone seriously wants to try to replace git they need to understand how git works first at the level of an advanced expert first anyway. At that point, you can have the discussion about how to make git, with better LFS/monorepo support, with a more pleasant UI/UX, etc.
There's a lot of "useless" complexity in the Git UI. The naming of the commands and the underlying concepts are quite inconsistent. The commands themselves are inconsistent. It's rare to see somebody, even git fans, to dispute that the UI is far from optimal. And the warts in the UI are being improved.
These UI warts start to become invisible when you get used to them, but e.g. when teaching git they become painfully obvious again.
The underlying git architecture is simple and it's the simplicity that makes git powerful. Powerful tools don't necessarily need complex interfaces. In fact power often comes from simple interfaces that compose well.
Totally agree on your main point though. The benefits of switching are far lower than the costs.
Once you understand this, GIT makes a whole lot more sense. The trouble with failed merges is an artifact of trying to appear like it tracks deltas. Merge conflicts when you're the only programmer on a product (but have multiple computers) are maddening.
I blew away .git and everything in it, restarting a repository, more than once because I couldn't resolve a merge conflict, before I learned that simple fact.
I've got from .zip files of my source code on floppy disks, to SVN, then Mercurial, and finally GIT. GIT is amazing, though the UI is a pain, as most agree.
But there is no machine-readable metadata that the commit is the same in both cases. In simple cases you can work around this by basing the patch off of the merge base but if there are conflicts that doesn't work well.
The result of this is that these two different but logically identically commits will cause future merge conflicts and make questions like "does this branch have this patch" much more difficult to answer.
This is something that I think https://pijul.org/ does quite well. Their base unit is a patch and that can be applied to multiple places. You can also have snapshots which are logically a collection of patches.
If you rename the file using git and that is the only change in your file, then it works:
git mv oldfile newfile
I might be misremembering the technical details, but isn't that only the case in a git repo with zero pack files?
Will grant that the lack of metadata on renames can be issue when a file is heavily refactored alongside it's relocation.
It can't display me the branch I'm working on, starting from the commit that began the branch, until the last commit. (Or if it can, I have no idea how to do it.)
This isn't a functionality issue, but rather a conceptual one. Git just fundamentally thinks of branches differently than I do. (And, maybe most humans?) To me, a branch begins ... when I branch it. To git, a branch begins ... at the very first commit ever in the repo, since they consider commits that were made before the branch point to still be part of the branch. The branch point then is just another commit to git - and not even one that they think is important enough to highlight. But without any ability to clearly identify the branch point, you can often find yourself looking through a forest of commits that are irrelevant to what you're really searching for.
I spent the better part of an hour the other day trying to figure out how to identify my branch point. (For the record, the best solution I eventually found was: cat .git/refs/heads/<branch name>.) Just my $0.02, of course, but IMO this is absurd - and a big feature gap / user friendliness issue with git compared to other VCS's.
IMO git is just another step in the evolution of VCS's, and not necessarily even one of the better ones. Its concepts, functionality, and feature set are focused primarily on distributed development and multiple people maintaining different source trees ... which is fine for the Linux kernel and other projects that heavily use that use case. But many/most projects don't work that way, and for them a centralized VCS is sufficient. I have no doubt that a better VCS will come along and replace git one day.
git log <base-branch>..<branch>
git rev-list --count <branch> ^<base-branch>
git log --graph --oneline --allThat being said the unfortunate truth is that many, many, many users of Git rely on the implementation as the API, directly reading and writing to all corners of the `.git` directory. So even if the core command could support something like this (or a wrapper could be provided) it would likely run into a lot of tooling issues.
https://learn.microsoft.com/en-us/previous-versions/azure/de... https://github.com/microsoft/scalar
The only downside is the per-file versioning, the update needs to be run on the top directory otherwise interesting problems happen.
Going from SVN to Mercurial was a night and day experience. Going from Mercurial to Git was a marginal improvement initially but a lasting change long-term.
Then there are the "visual" VCSes like Rational and TFS that are designed to _only_ work within an IDE and grokking them involves wading through hundreds of pages of corporate tech docs.
VCSes (or at least the ones I used) were generally awful before Git.
Git solves the deeply complex problem of distributed version control. Most complaints about git are actually complaints about the essential complexity of the problem, not the accidental complexity of the tool used to solve it.
I want to share 5k LOC with a colleague. Anyone who thinks that requires a decentralised solution in 2023 is unwell.
I recently wasted a few hours because I forgot to fetch before running some code. Sure the UI could be better (if you're very tired and you're told you're up-to-date you'll stupidly believe it), but fundamentally the 'feature' that caused this problem is something I'll never use. I'm always going to code with a connection to the internet. I shouldn't need to mentally keep track of local vs remote because that distinction simply shouldn't exist.
The underlying model is alien to many, it requires non black box approach to get out of pits sometimes but to me it's always worth knowing.
The only decision I dislike about git is the soldered staging area idea. There's a reason why most people end up stashing 109 times a day, and I think there's a golden idea in merging the two
Whatever effort you are putting into seeking or building an alternative, instead put that effort towards becoming a Git expert.
building an alternative naturally is more work than to learn git for one person, but not everyone needs to do that. a few people building an alternative could save the learning effort of many more people.
instead i'd like to point out a different reason why rebuilding git may not solve the problem:
git is good because it is powerful and flexible.
that power and flexibility by necessity makes git more complex and difficult to use.
you can build a simpler system, but then half of todays git users won't be able to use it because it won't have the features they need.
and a system that has all the features of git today and also solves the problems it currently has will be even more complex and less straigt forward to use.
the best approach is probably to have a common backend that has all the power and flexibility needed, and multiple different frontends that are simplified to only expose the features needed for a particular workflow.
I'm old enough to have used CVS, SVN and then a bit of git.
I remember the big selling point to go from CVS to SNV was "you can put binaries in there!" ... which we never did.
And now I see the selling point to go from SNV to Git is "you can do really complicated stuff like rebase".... which we've never done.
For a team who just does run of the mill distributed version control stuff (commits, diffs, blame, conflict resolution), can you explain why git is such a huge benefit over SVN?
Git really needs a good git tutor game.
Like vim tutor or any one of the n*x/shell games.
For example, git's cli user interface is monstrous (yes, I know, you personally have 800 cli commands memorized and get them all right every time, that doesn't make it "good"). From the outset, the maintainers of git basically decided "it's too much work to make all the cli flags behave and interact consistently" so they didn't. This allowed git to grow fast, at the cost of the cli user experience.
That said, git is big enough that multiple companies have come along and "solved" the git UI problem. None of these aftermarket UI layers are perfect, but there are enough of them and they are different enough that you can probably find one that is good enough for you, along whatever axis you personally dislike the git UI (examples include [0], [1], [2], which tackle very different user workflow problems).
I drop to the CLI occasionally, especially for multi step or automated scripts, but you can pry a nice visual commit graph and full featured integrated diff viewer from my dead hands. GIT is powerful and option-laden; the perfect tool for a UI to aide in discoverability. The CLI feels like programming in a text editor vs a real IDE
IDEs and text editors sometimes have nice Git integrations in the UI, but I wanted standalone software that I can use for anything from various programming projects, to something like gamedev projects (with Git LFS) or arbitrary documents.
In the end, I just forked over some money for GitKraken, it's pretty good, especially with multiple accounts on the same platforms, when you want to switch between them easily: https://www.gitkraken.com/
There's also Sourcetree which I used before then, kind of sluggish but feature complete: https://www.sourcetreeapp.com/
For something more lightweight, I also enjoyed Git Cola on various OSes: https://git-cola.github.io/ Even Git documentation has a page on the software out there, a good deal of which is free and has good platform support: https://git-scm.com/downloads/guis
Quite frankly, I spend like 90% of the time using a GUI interface nowadays, when I want to easily merge things, or include very specific code blocks across multiple files in a commit, or handle most of the other common operations. Of course, sometimes there's a need to drop down to the CLI, but you're right that some GUI software feels like it actually improves the usability here.
That’s probably really all there is to these discussions, good old technocratic chauvinism :)
I really wish the 4.* release supported a dark theme, it's the only thing keeping me on the 3.* release, and I dread the day I'll have to switch for whatever reason...
[0] https://github.com/martinvonz/jj
[1] 4 montsh ago, 261 comments https://news.ycombinator.com/item?id=36952796
[2] 2 years ago, 228 comments https://news.ycombinator.com/item?id=30398662
Perhaps the author can explain - when you clone the repo with
jj git clone
It pulls down a branch that is auto generated like pull-(hash)
I can’t understand how not to get that corrupted so when I do a jj log, I get very weird branches or heads or I’m not sure what.
Another way to say it is that everything works great until I have to pull down the repo from another machine - then the branch history is not what I expect. And I just couldn’t make sense of it or get it to square with what I expected.
I actually created a custom GPT and fed it the jj code and documentation to try and get it to explain it to me to no avail. Jj is so good, I’m willing to give up IDE integration with git if I could just crack this nut.
Even though Python is the language I’ve used most, I wouldn’t want to use it for something with a lot of uncertainty and that will suffer many changes. Type systems make it so much easier to change things early on without breaking everything. I’d probably pick F# or similar.
Most of the developers (some of them being former Mercurial and Git developers) including me generally seem to like it. Based on my own experience, I think it's a pretty excellent choice, but I'd be a bit biased as a die-hard Haskell/C programmer for something fast with types.
One of the most amazing things about Fossil is how you can track the history of a file not just backwards, but also forwards, something which is pretty whacky with git.
_Nothing_ history-relevant can be changed via manipulation of the Fossil db. In terms of db records, as opposed to space, the db is about 80-90% a transient cache of data which is generated from the remaining (100% immutable) data. Any changes you make to that transient data will be lost the next time that cache is discarded and rebuilt. A longer explanation can be found at:
https://fossil-scm.org/home/doc/trunk/www/fossil-is-not-rela...
Unless of course you spend time to learn git, but its complexity is closing up to C++. And using a VCS shouldn't require that amount of effort. It should just get out the way (I must admit, git usually gets out of the way, as long as you use only the base commands... but when it gets in the way, that's when the fun starts)
There's differences, such as that pgp is more complicated under the hood and it being a cryptographic system that needs to be foolproof whereas in git you can nuke and re-clone without data loss most of the time, let alone confidentiality/integrity loss. It just feels very similar in that only expert users properly use it and most people who could make use of it don't bother learning because the interfaces available are such a struggle (beyond basic operations anyway)
Whether it can all be solved with a simpler user interface, or whether it would require a simpler underlying system to be able to make simpler standard operations, is where I'm not sure
(Pretty sure I've said this before and it's been shot down before, so...)
* UX, obviously.
* Large files (LFS is a not-very-good poorly integrated hack)
* Very large projects (big company codebases). Poor support for sparse/partial checkouts, stateful operations (e.g. git status still scans the whole repo every time on Linux), poor & buggy support for submodules.
* Conflict resolution. It's about as basic as it can be. E.g. even zdiff3 doesn't give you quite enough information to resolve some conflicts (you want the diff for the change that introduced the conflict). The diff algorithms are all fast but dumb. Patch based VCS systems (Darcs, Pijul) are apparently better here.
IMO the most interesting projects that are trying to solve any of these (but not all of them sadly) are Jujitsu and Pijul.
Isnt one of git's core features that it can work as a patch based system?
It's my understanding (and please correct me if I'm wrong) that Linux patches can come in via mailing list, as a diff. That would make the person committing different from the owner of the change (also reflected in git's design)? Do Darcs and Pijul just have a string of patches on top of the original source file?
It uses an algorithm which is great from a computer science point of view (low algorithmic complexity, minimal length, etc.) but pretty bad from a semantic point of view (splitting up blocks, etc.).
There are a couple of attempts to improve this (DiffSitter, Difftastic) but they don't integrate with GUIs yet.
LFS is ok but it still feels like a kludge to me.
The cli does not bother me, there are many tools that offer alternatives/overlays/UIs. Not saying it is perfect or even good, but it's good enough - for me at least.
honestly keeping the large binaries as loose objects would be fine except for performance. which should be something that could be improved with cow filesystems (lfs does use this, but limited by what git's implementation can support)
or it may be enough to incrementally improve lfs, i do see that ssh support is showing up which might help a little. need to do something about smudge/clean filters too, which would require new support in git itself.
Instead everyone switched to a "distributed" version control system that is such a pain in the ass it is all now hosted by a single company.
Kids these days can't even code at all without chat-GPT, there's a central server hosting the git repo anyway, the whole architecture feels like it was designed for dial-up.
I can't think of anything that was doable 30 years ago compute and bandwidth-wise, that we can't do today due to performance reasons, except client-server source control...
My colleagues evaluated git and thought it was too complicated. A few years and a lot of employments later, they’re all using git. I don’t know if it was peer pressure from enough young recruits, but the verdict is clear: git is better than SVN.
A github server is much easier to set up than a subversion server. The reason people use github is because it's free, and because it has issue tracking and a wiki and forking which plain git knows nothing about.
And in general I feel that their gh cli tool doesn’t get enough praise. Being able to do easy API calls and queries for use in shell scripts (or just the terminal) is great, and the gh copilot is occasionally useful as a refresher for command syntax, or for deciphering some oddball git command you found online.
It’s a massive beast to tackle, and few people have a reason to. I don’t see anything doing what git does having a chance at competing with it. It requires a paradigm shift and a completely new product/approach to versioning to break the git dominance.
You can't set up a GitHub server.
Oh here we go again. You are wrong my friend. Firing up a basic SVN server is a matter of minutes. You just need to run several simple commands.
My first assignment was to spend several days picking apart an extremely nasty merge conflict from two branches nearly six months diverged. That was a very stupid thing to trust me with, as a major refactor was being blessed by my idiot hands.
Management could not figure out how they wanted to maintain a master/production branch.
Our 'trunk' was the develop branch, and any time we wanted to push to production.... We deleted the master branch and made a copy of develop. Master branch had no history, you had to track it back into develop and hope you found a trailhead from there.
It was a very bad time, and we were left with a very bad product. By the time I left, the codebase was so rotten and broken that we'd abandoned all hope of fixing the deeper issues.
I really hated subversion, but mostly the company was just unbelievably mismanaged. I'm sure you can use SVN in a sane way, just not like this
I do agree with your point about different IDEs, but I think VSC is actually one of the better tools for helping development teams unify the way they code. Then again, all our developers use VSC, except for that one guy who is stuck in regular VS and often has a ton of swearing because of it. Which is pretty much the legacy experience of regular VS from having used it for 10 years myself. It still amazes me just how slow it becomes with certain plugins, and how bloaty the various templates are. But hey, he’s happy/angry with it sooo.
I guess the story becomes a little different if you work with PHP or Java or similar, where VSC is arguably much worse than its competition, but you’ll likely still have some developers who prefer it because they also work with other languages. Or in Python heavy environments where there are also a lot of great IDEs for the more ML/AI/BI side of things.
Examples: - Book author using markdown / static site generator to publish a book. Uses visual editors like Typora. - Product designers for open-source hardware. Various design files, SVG etc.
I’ve experimented with a “GUI only” git flow - just to see what is possible, so I could introduce the concept to others.
I found GitHub desktop app (https://desktop.github.com/)did a great job of visually showing git flows and functions, but for a non-tech/programmming person, the tool would be daunting.
Curiosity what your suggested tech stack would be - sans Terminal…
Fossil views the historical record as immutable. Your sequences of mistakes, iteration, failed experiments are all diligently recorded. I like that, means I get to revisit missteps and abandoned branches later. However it is clearly a scaling hazard. I don't want to see the incremental hacking around of a thousand other people. Nor would I want to prohibit that sort of exploration.
My personal work is in fossil repos, going back a decade across various operating systems and versions of fossil. It has literally never let me down.
If the VCS would have an understanding of not only what has changed but also how this affects the code, it could deduce a lot if interesting facts about commit blocks.
Like ignoring simple refactorings (e.g. renamings), reducing merge conflicts, etc.
Pondering it, most of the easy things I can think about are really workflow issues on top of git. Git doesn’t exactly enforce them all though so maybe tighter integration would be a reason to change from git if it could not be adapted. Short of that, it’s hard to imagine that a new generation of engineers simply won’t do a new thing to do a new thing; there will be a “git considered harmful” article or a “modern” replacement for git.
As for DVCS, the best one I've used is Darcs: https://darcs.net/ There are some sticky wickets (specifically, exponential-time conflict resolution) that hindered its adoption.
Thankfully, there's Pijul, which is like Darcs but a) solves that problem; and b) is written in Rust! The perfect DVCS, probably! https://pijul.org/
Speaking of your analogy, the role that most software developers fullfil is not an engineer wondering about the oscilloscope, but rather the construction worker installing electrical fixtures wondering why the cable clamp has such a weird interface. Both the oscilloscope engineer in an office and the worker doing the field work would benefit from having a simple and reliable tool fit for purpose of cable clamping.
There is certainly a need for competent and "proper" software engineering that require special tools and detailed training, but I would argue it's niche and filled by people who build the tools themselves.
IMO the largest share of developers today are doing brick-laying work (which of course takes skill, I am not underestimating it) and would benefit a lot from having simpler tools - they don't need to know how to use an oscilloscope at all.
Git definitely has some warts but you are right. It is an industry tool for expert, professional use. Some complexity is inherent to the problem of version control.
Learning how to use your tools is part of ANY trade.
You can have both - powerful and user friendly.
This idea that engineer's tools must be a mess that is fine as long as enables to do something is idiotic.
The same argument was repeated whenever C or C++ vs Rust discussions were happening
"Just learn C and memory management (and all the quirks)"
"Just use this new language constructs and you're fine..."
and in reality we ended with a lot of CVEs - 70% in both Chrome and Windows were related to mem. issues.
There's absolutely no reason why git's CLI cannot be better than it currently is. Once again - there is no reason.
Proof? There are CLI wrappers or even GUIs like GitHub Desktop that make whole experience way better.
it's just backup of current state with irrelevant commit message. Everything is described at the end of the work in PR's description and squash merged.
I think they just want to replace some of the words with alternatives that they prefer. Because at some point someone is going to winge that update should be syncronise and not pull, and save should be push and therefore git is the worst.
Git is distributed, and that means you can't get away from push, pull and fetch, however you name them.
If want you want is a way to avoid making "New New Presentation FINAL 2", then pretty much all features of most source control systems are superfluous.
To me that doesn't mean Git needs fixing, it means it's definitely not the right tool for your job.
Besides that it's pretty much endgame in my opinion if you consider only the functionality it's meant to solve. If another "better" VCS would ever become popular I feel it would have to be a drastic change to the way of working with VCS, even more drastic than SVN to Git was. There's some cruft in Git that could probably be taken away, and that would make Git better in a theoretical sense, but in the real world that would never happen (unless we get sideswiped by another industry or platform).
The way we work around that missing feature is by tagging commits so we don't forget what revision a release was made at for example. A sequence of release tags basically is a meta branch, a history of the release branch, but managed manually instead of through git.
While git is good under the hood, then it has not really user-friendly interface.
Also git heavily benefits from GitHub's success, which locks us with git :(
I wrote about it here https://trolololo.xyz/github - GitHub is really good, but there's a small problem with that
A lot of alternative tools come up because of people writing them being unwilling to learn git. There are a handful of concepts and a few handfuls of commands and thats it.
And once someone learns git throroughly, they usually come to see that it is actually good enough, and dont bother making something new.
They don't even bother to add directory tracking ;)
I'm sure git can be improved, but I think the biggest improvement comes from the user being improved with their understanding of the scope of capabilities. I have yet to see a good tutorial on this (among the plethora of git tutorials out there). This reminds me of the (excellent) video where the "Harvard Professor Explains Algorithms in 5 Levels of Difficulty" [1]
[1] https://www.youtube.com/watch?v=fkIvmfqX-t0
I would love to see a Channel where that is the entire theme - explaining everything in 5 levels of difficulty.
All this being said, at each of the "5 levels of difficulty" of git, are there improvements to be made. I'm sure there are. It would be good to focus the answer on each of those levels.
I think git escaped to the public before having a designer's refinement. Users need to learn about the glue and speak the glue language to git to make it work.
I think there's no magic and we'll still have to resolve merge conflicts on our own, but my sense is it does simplify repetitive operations.
Hope this helps!
For standard needs, probably going to be difficult.
You're mostly going to hear from people here who are annoyed with Git or otherwise more interested in the topic of version control than the median developer. For me, I think it provides a quite robust and well thought-out set of primitives, and then composes them upwards in ways which are about as good as one can expect.
Some stuff obviously isn't well supported. Using the same Git repo to hold large binaries as well as source code is not well supported unless you reach for LFS - that's the biggest downside I see.
Fossil would be my next bet. I'm waiting for someone to make an archaeology.co to rival GitHub.com for it.
The existence of the staging area is a poorly thought out part of the design. No other VCS uses it because it was a bad idea that makes the simple case of commuting changes more complicated.
>Fossil would be my next bet. I'm waiting for someone to make an archaeology.co to rival GitHub.com for it.
Which is exactly why fossil will not be the next big VCS. Ignoring all of the projects on Github add forcing people to move to a less featureful, less integrated, less familial forge just to use a new source control system is a hard sell. The approach of Sapling and Jujutsu where they support the git protocol so that they can be used Github will make them much easier to adopt since it can happen incremenetally and it fully replace git for people.
IMHO the next VCS model should follow a centralized-first, decentralized optional model. Which would be a flip of the decentralized-first model of git.
I also think GitHub is in a unique space to really innovate on git and it’s a shame they’re not.
For example, I shouldn’t need to make a fork to make a PR. That’s absurd and the GitHub server should be able to apply ACLs based on the push identity.
There’s a couple more of these suggestions I can think of, but yeah, GitHub should do more in this space.
> the GitHub server should be able to apply ACLs based on the push identity
That’s essentially exactly what a GitHub fork _is_ – an ACL’d set of refs you’re allowed to control, separate from upstream’s set of refs. I guess – what would you have us do differently?
Disclosure, I work for GitHub and am the product manager for Git stuff.
What really slows me down, and honestly just kinda annoys me is that when I’ve cloned some upstream repository onto my machine and made a bug fix, why do I need to then fork the repository once more and push the commit to my fork first and then go through the theatrics of making a PR?
GitHub the server could, for example, fake my branch on the origin/upstream. It doesn’t need to actually make the branch on the upstream, and can smartly make a volatile fork, and prepare the changes to become a PR.
Basically, since the server knows my identity, and knows if I can or can’t make a branch on the upstream repo, it can handle the boilerplate of making a fork for me.
What I want to see from GitHub is embracing the power of there being an authoritative server between the developer and the repository.
But yes, I think we can.
Almost everything can be better:
* Merges/rebase/branch management.
* Project management.
* Diff.
* Large files.
* User management.
* Partial checkouts.
* ACID semantics.
* Binary file management.
* User experience, including making it accessible to non-technical folks.
* A bunch of others.
The hard part about it is that it's different for every language, so to support it for a language, you have to implement diff for that language.
I'm hoping to make money off of companies wanting those languages.
Of course, as long as you only do additive changes via commit/merge/revert, the logical history is equivalent to the physical history, but commands like rebase break this model. And despite the flaws of rebase workflows, sometimes it is the best option, like when maintaining a fork.
To my surprise Vim actually has something like this - logical history with undo/redo and physical history with g+/g-/:earlier/:later
Another thing I would like is some way to "fold" multiple small commits into one bigger one (for display purposes only) as it would let me split large diffs into minimal, self-contained commits while maintaining a reasonable git history.
thanks in advance
The complex explanation: a git repository at a particular time consists (mostly) of a graph of commits. This graph represents the logical history of code changes in the repository. The graph can be updated in an append-only fashion (using commit/merge/revert) or in a destructive way like with rebase and reset. The physical history is simply the history of the graph over time.
I (briefly) looked into using git as an alternative to PDM for solidworks CAD files and it turns out git is absolutely not a good fit for this use case. Not surprisingly, I mean it wasn’t designed for that at all.
Point remains though that the world needs a better version control system, something like git, but that works with non text files, because boy do the actual solutions I’ve tried suck compared to the git experience
Software engineers are so lucky to have such a powerful tool for free, mechanical engineers or their companies pay tens of thousands of dollars for version control software that is far worse
PDM is a bit more than just version control but the version control is what my company wants and it’s so painful
The best UI in my opinion is sourcetree, which is not available on linux.
I worked with sourcetree years ago (switched to linux last years and used smartgit client), I don't know the current state, but old versions are available to download, and don't require an atlassian account.
Some improvements I could suggest are the ability to ammend a commit which is not the last (and not pushed of course).
Currently if you want to ammend the third last commit for example, you have to soft reset last commit, push it to stash, reset the new last commit again and push it to stash, ammend the changes, then pop the previous 2 stashes and commit them one by one. This could be easily automated.
For a new versioning system we do not need twenty different choices. We need one free, open, and solid solution that everybody uses.
What the main leaders of the industry should really is to found a groupo that defines that standard. This would be their chance to really make the world a (slightly) better place.
It wasn't until 2011 that Subversion dropped below 50% market share in the Eclipse Community Survey. Something new and shiny will come along and replace git.
Similarly, the solution to git is a subset of git (strict git).
Gits problem is that it is too powerful and assumes that it's users are all git experts. You should be able to run git in 'easy-mode'. Add, commit, checkout new branch, revert, squash merge. That's all 90% of people need.
Then the intermediate folks can run it in mode 2, adding the ability to rebase, reset heads, cherry pick, revert, stash etc. This covers the next 9%.
The last 1% can use it in mode 3 for the rest.
Once you take away the fear of what you could break, git becomes far less intimidating.
Git is much harder to use than SVN. Particularly you see engineers struggling with resolving conflicts.
One benefit of Git was decentralization, but now Git is Github, it is centralized again.
Builds, IDE, programming languages should be 1st-class citizens. I don’t want to wrestle with .gitattributes or .gitignore.
What would I make better?
- Fast with large files. - Simpler. - Improved commit meta-data.
Many of these datasets have many many images, videos, audio files, text as well as structured tabular datasets that git or git-lfs just falls flat on.
Would love anyone to kick the tires on it and let us know what you think:
https://github.com/Oxen-AI/oxen-release
The commands are mirrored after git so it is easy to learn, but optimized under the hood for larger datasets.
Git won by being free and since a lot of people think using the terminal is somehow better, but we could have done a lot better in terms of UX and power. Command line programs force you to have a lot of context in your mind, and if git would have been designed with a UI it would have been probably better for everyone.
I still hope for better tools, but they'll probably be based on Git, just with better default flows.
I suspect most people use just a tiny subset of git day-to-day, and google the rest when it comes up.
For this reason, I think if git is replaced, it won't be because whatever comes along will be better objectively, it will be because a few reasons are touted that most people don't understand but are willing to repeat, and some momentum builds behind the alternative.
The main innovations seem to be:
- The concept of a "stack" which is somewhere between a commit and a branch, a group of commits.
- Better UI, especially for Github notifications.
The end result is he feels safer using advanced git features and can move faster, especially when working within a team of multiple devs.
That seems to be just some conventions for branch naming & rebasing. The most value seems to be in Graphite syncing that state into Github PRs.
It says a lot about these people (e.g. narrator of this video) that they think this is something novel. The Linux kernel community has done "stacked commits" even before Git existed..
Unless it's about webdev & on youtube, it doesn't exist?
I use a wiki which internally uses RCS, but you never see it. The only reason I even know is that I needed to scan older versions of some assets and it was straightforward compared to what you'd expect with Git. (Other bonus, attachments and meta pages are stored as actual files. With a little bit of code you can cobble together an automated page builder for e.g. physical assets.)
I consider rsync --link-dest a version control system.
But if you are doing binaries, because you are an artist/do 3d modeling, you probably sill use svn.
And I am still checking in on https://pijul.org/ from time to time
There's work around a bunch of languages that are not based on text, some have their own editor or a tool to manage a canonical representation in text for you that would make them friendlier to git.
- https://github.com/yairchu/awesome-structure-editors/blob/main/README.mdThe current trend (most popular and IMO likely to succeed) is to make tools (“layers”) which work on top of Git, like more intuitive UI/patterns (https://github.com/jesseduffield/lazygit, https://github.com/arxanas/git-branchless) and smart merge resolvers (https://github.com/Symbolk/IntelliMerge, https://docs.plasticscm.com/semanticmerge/how-to-configure/s...). Git it so flexible, even things that it handles terribly by default, it handles fine with layers: e.g., large binary files via git-lfs (https://git-lfs.com) and merge conflicts in non-textual files by custom merge resolvers like Unity’s (https://flashg.github.io/GitMerge-for-Unity/).
Perhaps in the future, almost everyone will keep using Git at the core, but have so many layers to make it more intuitive and provide better merges, that what they’re using barely resembles Git at all. This flexibility and the fact that nearly everything is designed for Git and integrates with Git, are why I doubt it’s ever going away.
Some alternatives for thought:
- pijul (https://pijul.org), a completely different VCS which allegedly has better merges/rebases. In beta, but I rarely hear about it nowadays and have heard more bad than good. I don’t think we can implement this alternate rebases in Git, but maybe we don’t need to; even after reading the website, I don’t understand why pijul’s merges are better, and in particular I can’t think of a concrete example nor does pijul provide one.
- Unison (https://www.unison-lang.org). This isn’t a VCS, but a language with a radical approach to code representation: instead of code being text stored in files, code is ASTs referenced by hash and stored in essentially a database. Among other advantages, the main one is that you can rename symbols and they will automatically propagate to dependencies, because the symbols are referenced by their hash instead of their name. I believe this automatic renaming will be common in the future, whether it’s implemented by a layer on top of Git or alternate code representation like Unison (to be clear, Unison’s codebases are designed to work with Git, and the Unison project itself is stored in Git repos).
- SVN, the other widespread VCS. Google or ask ChatGPT “Git vs SVN” and you’ll get answers like this (https://www.linode.com/docs/guides/svn-vs-git/, https://stackoverflow.com/a/875). Basically, SVN is easier to understand and handles large files better, Git is decentralized and more popular. But what about the differences which can’t be resolved by layers, like lazygit for intuition and git-lfs for large files? It seems to me like even companies with centralized private repositories use Git, meaning Git will probably win in the long term, but I don’t work at those companies so I don’t really know.
- Mercurial and Fossil, the other widespread VCSs. It seems these are more similar to Git and the main differences are in the low-level implementation (https://stackoverflow.com/a/892688, https://fossil-scm.org/home/doc/trunk/www/fossil-v-git.wiki#....). It actually seems like most people prefer Mercurial and Fossil over Git and would use them if they had the same popularity, or at least if they had Git’s popularity and Git had Mercury or Fossil’s. But again, these VCSs are so similar that with layers, you can probably create a Git experience which has their advantages and almost copies their UI.
(Another plausibly correct answer of course would be "AXBGAXB", but again `recursive` doesn't know enough to guess this answer.)
What features do you think need to be improved? From a purely UX pov I think git is probably the best software I’ve used. It just works.
Git as practiced by GitHub and gitlab is awful quite a lot of the time.
If it doesn't have multiple clouds providers with pull requests, and at least one of those clouds providers isn't a megacorp, it probably won't be the safe boring choice, unless it's fully P2P.
It needs to be packaged in distros, have GUI integrations, etc.
Fossil looks really cool, I really like how they integrate the wikis and issues. But I don't know anyone who uses it, and the cloud providers seem to be smaller "Could go away any time" companies.
I've never really explored any other VCSes, because none ever seem like they're going to be serious competitors.
I'd be more interested in enhancing Git, but it seems a lot of the most interesting git plugins and extensions aren't updated, like GitTorrent.
There are some GithHub-alikes, the most obvious is GitLab which you can also host yourself but all (or at least some of) the extras you get for free with GitHub are behind payment walls.
My current favourite is Codeberg, it uses Forgejo underneath (which is a fork of Gitea itself a fork of Gogs - all of which you can self host). Codeberg is run by a non-profit and are very much aligned with my ideals. They are also slowly adding nice features like their Woodpecker CI.
One that is growing in popularity and is a little less "GitHub-y" is SourceHut (which also has Mercurial support).
The main issue is that GitHub has really cornered the market. They give so much out for free that is difficult for others to compete with and it has become the de-facto place to host your project. This can mean that hosting anywhere other than GitHub will limit discoverability and contributions from people if they don't want to make an account or work out how to deal with whatever forge you are using.
However one thing that is coming that may help alleviate some of that is forge federation which will allow you to interact with various forges from your "home" forge - which hopefully prevents the need to make an account to make PRs or raise issues.
Edit: I see your other comment now, what could a better GitHub be that supports a better-than-git VCS. Well there did used to be places to host Darcs projects like the Darcs Hub but I don't know if some of the newers ones like Pijul or Jujutsu have any forge support yet.
Edit2: Oh it seems Pijul has "The Nest" for hosting.
There's no native forge yet.
¿Que?
If you're wondering whether we can make something better than GitHub, there's dozens of git hosting alternatives that you might like better such as Forgejo and GitLab.
If you're saying "but I already know git", then there's still dozens of alternative hosting sites or methods!
There are others that are almost as good, I suppose what I should have said is "Can we make something better than GitHub for the hypothetical better non-git VCS", since without that it's hard to imagine using anything but Git.
A lot of proposal in this thread would have improved one of these at huge detriment of others.
The only way I could see it changing is if we have a complete paradigm shift. This is what happened when we went from SVN to Git (centralised to distributed).
Hey, git don't even store diffs at first place. All it store is the snapshot of all directory content. The diff git showed to you is actually generated on the fly. You can even just clone a single commit and get the content out of it if you are doing ci and don't need the complete history.
But you could also do better than Linux or OSX as an OS, but the mindshare and industry investment there is so strong, moving to a new thing is a monumental undertaking
sapling scm from meta has I think the best cli and VS code UX https://sapling-scm.com/
jj from google is also mercurial derived with very similar cli features like histedit and has support for deferring conflict resolution https://github.com/martinvonz/jj
There is definately scope for a beginner friendly UI/UX. Julia Evans has a post lately about confusing aspect of git. Ability to version control large files (like git-lfs) would be a nice addition.
I've read so many comments to the effect that "X is a solved problem" when it clearly wasn't that I've come to conclude that the phrase means the opposite of its surface value...
I'm pretty sure that Git is not the end-of-line as far as VCSs go. Whether I will ever change my VCS, again, ever, that is an entirely different question. I've been through so many of them (RCS; a bit of CVS; a bit of Subversion which promised to be CVS-without-the-flaws, which it was not; Mercurial, because hey it was written in Python, so must be good, right? right?; finally Git; and of course the usual `report-v4-revision-good-one.doc` renaming game; plus a plethora of backup solutions, some really bad ones among them)—so many of them I'd be loathe to switch to yet another one, except maybe Fossil, which I almost did.
So yeah, I had totally forgotten about https://fossil-scm.org ; the reason it didn't become my go-to solution is probably mainly the fault of github.com, which I find too good to be ignored, and I don't want an 'impedance mismatch' between my system and their system. But maybe it would be doable and maybe Fossil is good enough to be worth it; at any rate, go read their docs, especially where they compare themselves directly to Git and give lots of good reasons for their way of doing things. This is the same people who are doing SQLite, so I'd say a trustworthy source of reliably top quality software.
Other than that, my personal way of dealing with the complexities of Git is to avoid using parts that I don't need or don't know about well enough (i.e. almost all of it). I use it to check in changes, give one line of comment, and upload to github.com; then of course cloning and updating existing repos as well as setting up a new ones is within my skill set. Branching and merging not so much. So it's like `git` plus `clone`, `add`, `commit`, `push`, that's all; also, I use Sublime Merge for part of these (reviewing changes, adding related changed chunks, commit) which I must recommend for the piece of fine software that it is.
I also at some point toyed with gitless (I think it was called) which promised to be Git, but simpler for the simple things and you can always fall back to Git proper where called for; this I find a good proposition that I like (Markdown: fall back to HTML; CoffeeScript: fall back to JavaScript) but somehow gitless didn't stick with me; I guess that's b/c I've already tempered down my usage of Git to a point near absolute zero, and command line history + Sublime Merge does the rest.
So give your contributors developer accounts on your Fossil instance, which is super-cheap to set up, being a single binary with nearly zero external dependencies. (Those being OpenSSL and zlib, which are table stakes these days.) My containerized build is a single static binary that compresses to ~3.5 MB, total, all-in.
If you're concerned over the lost promise of easy PRs from randos on the Internet, I question your premise. My experience is that below a certain project popularity level, there is less than one total full-time developer on the project, even counting all possible external committers. Below this threshold, why optimize for external contributors? If someone has a sufficiently valuable patch, they can deal with getting a Fossil repo login or sending a patch.
I've been the maintainer of a piece of software for coming on two decades that's in all popular package repos and have _never_ gotten a worthwhile PR for it via GitHub. I spend more time using their code commenting features explaining why this, this, and this make the change unacceptable, after which the PR submitter goes away rather than fix their patch. It's a total waste of time.
I did once upon a time get high-quality external contributions, but that was back when the project was hosted by Subversion, and it didn't matter that posting patches required more work than firing off a GH PR. People who have sufficient value to commit to a project will put up with a certain level of ceremony to get their code into the upstream project.
(To be fair, I expect the reason for the lack of quality external contributions is that the project is in some sense "done" now, needing only the occasional fix to track platform changes.)
If you are lucky enough to have an audience of outsiders who will provide quality contributions, Fossil does have a superior option for patches than unified diffs. See its "patch" and "bundle" commands. This lets your outsiders send a full branch of commits with comments, file renames/deletions, etc.
…kind of like a PR. :)
If you absolutely require integration with Git-based tooling, Fossil makes it easy to mirror your repo to GitHub, which you can treat as read-only.
Well, I can't really say that that's why, but yeah. Mercurial's pretty great.
What matters is if git is good enough.
Or more specifically, is if git good enough for X when X is something that is actually being done.
I mean, git is good enough for my very minimal needs and the needs of people with much more sophisticated needs than mine (e.g. the Linux team). And since I know more about git than any other VCS (in part because there are better resources for learning git than any other VCS) learning another VCS for the sake of learning another VCS wouldn't help me get anything done.
None of that means git is good enough for your needs, but statistically, it probably is good enough for your needs because statistically, most difficulties with git are related to training and knowledge since the mathematics underpinning git are (to the best of my understanding) sound.
Which also implies (not accidentally) that being better than git requires better resources for learning the new VCS than git has, and that's a very tall order.
Good luck.
The most trivial example of a thing that is wrong with Git and which no amount of getting better with the tool can possibly help is "once you generate a conflict, you cannot perform any other versioning operations until you fix the conflict or revert". In particular, for example, you cannot commit a partial resolution of a conflict: you simply have to bail out and try and put your histories in a state that is more acceptable to the merge algorithms before trying again.