[jimbokun]

> Poetry does the same by default, using a format like >=1.23.4,<2.0.0. I find this less readable than ^1.23.4, but the effect is the same.

What???

I understood the first format instantly, but had no idea what the second meant until the author explained it.

[pie_flavor]

And then you look it up once, and now you know what it means forever. By contrast, the former expression is much wider with more going on, and furthermore you can't skim past it being sure nothing funny is going on because it may or may not be a range compatible with the latter form.

[scubbo]

> now you know what it means forever

Not, in fact, correct. Knowledge only cements itself in the brain when it's regularly referenced. Because `>=` and `<=` borrow well-established concepts well-established, they are both intuitive to people reading them for the first time, and easier to solidify or to re-infer for someone who's forgotten their meaning.

[deathanatos]

> Knowledge only cements itself in the brain when it's regularly referenced.

While true, this is a molehill, not a mountain, of a bar, like "coding once in a while". I'm doing mostly SRE work, and this syntax has no trouble sticking in my head, and I encounter it pretty regularly? (And heck, most of my work these days is in Python, so there I get the >=,< syntax and yearn for the ~mines~ caret, and I still recognize it?)

If you're actively developing a codebase, this definitely isn't going to be arcane trivia.

[saghm]

I'd argue that complaining about using combined "greater than" and "less than" operators instead of the caret is about the same size of molehill as complaining about the usage of it. Seeing either of them in dependencies at my job would be a pretty mundane event that I wouldn't bother trying to do anything about.

[saghm]

For what it's worth, even as someone who's known about the syntax for over a decade, I think there have already been two times this calendar year that I've mistakenly gotten the meanings of the caret and tilde operators on package versions mixed up and been corrected on it.

[j16sdiz]

No.

I have look it up years ago, and I don't remember all combination of `=` vs `^` vs `~` across all languages and package managers

[alexanderchr]

To make it even more interesting, in npm ^ works differently depending on what version you are on

[drcongo]

Also worth mentioning that a carat doesn't require quoting in your shell so you can type `uv add something^1.5` which is a hell of a lot easier to write than `uv add "something>=1.5,<2.0"`.

[DangitBobby]

I've looked up the ^ and ~ syntax differences probably half a dozen times. They are arbitrary and un-intuitive.

[jim33442]

I saw that ^ so many times in npm and never knew what it meant until now

[saghm]

I don't know if there's a good source for where this convention started but it's not that uncommon; offhand, both npm[1] and Cargo[2] support that format.

[1]: https://docs.npmjs.com/about-semantic-versioning#using-seman... [2]: https://doc.rust-lang.org/cargo/reference/specifying-depende...

[kjmr]

Once you know what the ^ means (I always think of “roof”), I do think that one character is easier to read than >=,<

[skeledrew]

Same. I found that wild to take in.