Hacker News Reader

Comment by jakobnissen | Hacker News Reader

Obviously the meaning of "Julia is performant" is that Julia CAN be performant in a way that Python can't. Of course it doesn't mean Julia will write fast code for you, how could it possibly mean that?

Edit: This is like saying people has to stop saying Ferraris are fast, because you can still drive slow in them

View on HN

No it is not. It's like saying a Ferrari in third gear will be faster than a Fiat in third gear, because Ferrari. Clearly this is not the same as saying that "therefore a Ferrari is as good as a Fiat", because obviously Fiat only goes up to gear three whereas Ferrari goes up to 7. But chant the "Ferraris are always faster" mantra without this context, and eventually all but your most technical users will be the kind of user who sticks to the third gear that they're used to, and still expect a Ferrari to be fast.

You want your Ferrari to go fast? Drive it like a Ferrari.

Also, Python absolutely can be as performant as Julia, if not more. The point is that Julia has language-builtin features that were created with this performance in mind, whereas for Python you have to go explore the relevant package ecosystem and go down a numba/cython/pypy/etc route.

And, yes, if you bragged how a Fiat could never beat your Ferrari in a race, and you agreed to race me against my modified Fiat with external turbo engines and I beat the crap out of your Ferrari, your "well that wasn't really a Fiat" complain would be valid, but only so much. Especially if Fiats were super-moddable by nature.

Eagerly awaiting matrix libraries written in pure Python that outperform BLAS.

Waiting for Octavian to replace https://github.com/JuliaLang/julia/blob/master/stdlib/Linear...

The fact that there is a competitive candidate replacement native in Julia (Although a few man years in engineering time wouldn't hurt) is telling.

"written in pure python" is just moving the goalposts.

There's no need for them to be "written in pure python". They just need to be seemlessly usable from within it.

Your argument is that “Python can call fast code written in other languages”? That is true, of course, but if that’s your definition of Python being fast, then it makes the notion of one language being faster than another meaningless.

You see, this kind of response is precisely the problem with the Julia community and the performance fallacies they present. What does this post have to do with Python? You claim to be the fastest language, and if someone questions it, you only mention how fast it can be compared to pure looping Python. It's a Machiavellian promotion strategy that has so far worked for Julia, but it isn't going to work any longer.

Hmm, I would be interested to see where exactly it is that these Julia advocates have claimed it to be "the fastest language"

Start with the false claim of JuliaCon21 banner, "Julia is the fastest high-level programming language", then go to the Julia Discourse and read all the false reports on the internet claiming 3X, 300X, 3000X faster than Fortran/C/Python/... Having a business to run does not justify making false claims.

The JuliaCon 2021 banner can be found here (https://juliacon.org/2021/). It says "JuliaCon 2021 (with JuMP-dev) was online and virtual 28th to 30th of July, 2021. Check out all the recorded talks on Youtube". A google search query for the phrase "Julia is the fastest high-level programming language" finds no hits.

https://www.google.com/search?q=%22Julia+is+the+fastest+high...

Please have some intellectual honesty or share some links as apparently you know where to find some things that Google does not know about.

Search the LinkedIn ads in summer 2021.

It turns out the claim is even is worse than what I remembered. It claims to be the fastest high-performance language. What a shameful Machiavellian act by JuliaComputing for lacking intellectual honesty. https://www.linkedin.com/events/juliacon2021-freeandvirtual6...

well the Julia version runs faster than rust once the bugs were fixed, and a slightly more optimized Julia version finished in under a second.

Reply on news.ycombinator.com