Pijul

@mgattozzi @kerckhove_ts I meant much better than 10 years ago ;-) Wasn't aware of the leadership crisis, thanks.

@pijul_org @kerckhove_ts I mean this is recently so I disagree the docs are better, nor is the tooling easy to work with, not to mention the leadership crisis regarding Eelco is enough reason to stay away.

Having terrible ideas like: what if I built a CI system that works the same locally and in the cloud? What if we didn't program them in YAML?

@mgattozzi @kerckhove_ts Its docs have improved a lot. The problem is, typing Nix is hard, also because Nix would need a really unintuitive type system. haskellforall.com/2022/03/the-ha…

@kerckhove_ts Yes it's not YAML but also it's the nix lang which has terrible docs and is frankly inscrutable. I really tried to enjoy nix, it fixes all my problems with software, but it was like stepping on glass. It's usability curve and sharp edges just aren't worth it.

@mgattozzi @nixos_org ?

Turns out I have other ideas like: what if we didn't program with Docker/Makefiles or use a language with actual docs and a not have leadership implosion problem happening in the past few months

@david_kinghorn @peterrhague Git is almost impossible to use beyond the first `git clone` if you think of it as diffs. This isn't an internal detail, it leaks everywhere.

@pijul_org @peterrhague My intention was not to describe how it works internally but a quick list of getting started commands. Git diff certainly makes it appear this way.

Literal skill issue. Git is a pain to learn but you’ve got to do it.

@peterrhague Not so clear. Another skill issue is to understand Git, but not enough that you can see how it slows you down.

@david_kinghorn @peterrhague Except this is wrong, but it would indeed make things so much better. Your can think in terms of diffs in Darcs and Pijul, but not with Git.

@peterrhague There aren’t that many commands - pull, merge, rebase, and branch - should get you able to do most things. But the key is to think of it as a tree containing object diffs rather than a file system. This should be taught in first year computer science IMHO.

@pedrohdev A gente esta construindo uma alternativa, mais fácil de usar e também mais adequada para repositories grandes.

@edvpml @hipsterelectron @yburyug You'll know the answer soon enough (sorry).

@pijul_org @hipsterelectron @yburyug Are you planning some kind of integration between the two systems? Or what kind of collaboration are we talking about?

we should all start trying to claim Homebrew on @teaprotocol and see how Max Howell likes it

@gony Many of these are totally objective, i.e. not "my opinion". Also, I wasn't the one who started that. I never "tangled" anyone else, ever. We have good relations with all other VCS maintainers/authors.

@pijul_org It's just your opinion. You can put it in your blog or something. Do not tangle others any more. Thanks.

Jujutsu(バージョン管理システム)のオタクなのでコミットをウォッチしているのだが、ドキュメントにPijul(別のバージョン管理システム)にインスパイアされましたみたいなことを書いていたら攻撃的だみたいに言われて消したらしい... オープンソースもいろいろあるのねえ... github.com/martinvonz/jj/…

@gony Even the current solution is dishonest: instead of doing a proper, objective comparison between different tools (which would have forced him to understand these tools), he chose to just delete references.

@gony Part of the Jujutsu marketing is plagiarised from Pijul (Martin claims to have "rediscovered independently" the phrase "first-class conflicts" after reading our website). Plagiarism is not respect.

@gony Another issue leading to that is that Martin and friends have been squatting our Zulip channel for months to "offer help/explanations on Jj", despite our repeated warnings that this was completely off-topic.

@gony Note that we suggested a better alternative, which was to learn about Pijul so they could do an objective comparison instead of saying Jujutsu is a strict improvement over Pijul (it's definitely not!)

@rrr00bb @yeedle @0xglitchbyte Have you tried Pijul? Sounds like no.

@pijul_org @yeedle @0xglitchbyte it is such a hard to use solution that nobody uses it. the usability solution isn't solved until you can just casually check in GB of artifacts without thinking about it.

Hot take: Universities shouldnt teach tools like Git Universities are meant to challenge thinking, dive deep into concepts from first principles, and push academic boundaries They are not there to show you how to use Git or any other tool Professors can make students aware of these tools, but it is not their job to teach them a tool they can learn in less than an hour

@goldenbergdev @TheGingerBill Also, none of this implies that Rust is the final language ever, nor that C is dead. Testing crazy algorithms and datastructures with top performance is where Rust shines the most IMHO.

@goldenbergdev @TheGingerBill On that particular example, there's even a famous paper in DB theory giving the wrong answer ("still wins" in the tweet above was a trap).

Mini-Rant: Rust was not built to be a fast language, it was built to be a memory safe language. Rust is compiled & that usually implies it will be faster than non-compiled languages. But that doesn't mean it is fast compared to other languages designed for performance.

@rrr00bb @yeedle @0xglitchbyte Wait, it's actually solved, we solved it.

@yeedle @0xglitchbyte large repos are still a bit of an unsolved problem. automating merges are unsolved. CRDTs, merkle trees, append-only data; it's the intersection of cryptography and functional programming in a lot of ways.

@goldenbergdev @TheGingerBill If the fastest way turns out to be SIMD, you probably don't want to think too long about how to make them portable, since you won't spend that time experimenting.

@goldenbergdev @TheGingerBill What's the fastest way to search through a sorted array of 512 values? Does it matter that these values don't fit in registers?