Another example of “I like my language and here’s why it is uniquely suitable for $thing” (here AI).
I just asked chatgpt today for research on whether any feature of a programming language which has been identified in studies to make it better suited for agentic coding. Or whether popularity in the data-set has been proven to matter or even cross-language learning (knowledge from one language benefitting the LLM in writing in another).
Aside from popularity, nothing I could dredge up could prove strong correlations with anything else - including types
Funny, I thought HN was about substance, not cheerleading.
If you have reason to downvote, present it.
Such as it is - I doubt it, I couldn’t find any papers, so either it goes against someone’s religion or feel free to present your proof where any of the traits of Rust the language has been proven to actually benefit LLM code generation.
Heh - good choices. Seems like a poorly reasoned article then.
Accepting the premise of 2000$ spent (I doubt it), the author completely forgets that the collectively saved amount far exceeds the 200$ of a single subscription.
Anyway, if I worked at Adobe, or any company with roughly the same business model, I definitely would feel nervous about now. And I still feel sympathy for the devs. But let’s be honest, Adobe principally benefits its shareholders and other dead weight, not so much the developers.
True. It's really cases like this that demonstrate the shortcomings of BSD/MIT licenses -- at least if you are of the position, that companies could and should contribute a sliver of profits back to the projects enabling that success.
(Much like contributing back in taxes for the infrastructure used and the educational system producing the qualified employees. But that may be a bridge too far for some of HNs readership :) )
Is it actually a shortcoming? People choose to use these licences. They explicitly say the companies can use it free with no obligation. Most are happy with that choice and say so. If they were not happy they can change the license for new code. They could use a license unpalatable to those companies and dual license with a paid alternative
Last week someone had left bags of apples left on their wall with a note asking people to take a bag. I guess their tree produced more than they wanted. I took one. I feel no obligation to pay them for it. An MIT license is the same thing applied to software.
One; you are a private individual- not the local shop which simultaneously picks up all free apples only to turn around and make money off of them.
Secondly, while we can’t speak to everyone picking any open source license - there are good examples of people complaining that they are stressed because they either are underfunded or having to maintain critical infrastructure pieces in their spare time - because companies aren’t donating in exchange for the free code.
You can say “don’t”, but the openssl, xz, … maintainers are often aware how much can break if they don’t keep at it - and so for a sense of societal duty they continue.
If your neighbor stops hanging bags of apples by their house it’s not like they have a reasonable expectation that someone is going to die from starvation or anything.
The comparison is valid because it is voluntary. If you give something away free you cannot complain if people take you at your word. Donating is an act of charity, not an act of exchange. I agree the companies should give back, but that is not how the world currently works.
What I do not understand is why projects that are short of money continue to use those licenses. Relicense to GPL 3 or AGPL 3 so small businesses and individuals can use it unhindered, and offer a paid license to those averse to that.
As xz demonstrated, maintaining for free out of a sense of obligation is not always good for society.
If you give something away free you cannot complain if people take you at your word
that's not the problem. the problem is that the people using the software then turn around and complain about bugs. but fixing bugs in a timely manner was not included in the free giveaway.
companies shouldn't give back because they get a benefit, they should give back because they expect continued maintenance.
projects aren't short on money because they give away their code for free. they are short on money because those users demand support without paying for that.
and they can't change the license because there is no license that allows anyone to modify and redistribute the code but can demand payment for maintenance. if they change the license people would stop using it, and the project would be dead. (or forked)
GPL 3 or AGPL do not help with this problem. it might stop a few big businesses, but again, it would stop them from using it, not motivate them to pay for it. (it might, now that they are already using it, but if that was the initial license , they would never have used it to begin with.
the only way to achieve what you want would be to start with a permissive license and then change the license later. but that is not workable. it would upset and alienate to many people and motivate some to fork the project. not to forget that once you have contributors, license changes are hard to do.
> that's not the problem. the problem is that the people using the software then turn around and complain about bugs. but fixing bugs in a timely manner was not included in the free giveaway.
Agreed, but there is no obligation on them to do it. Even if they feel a social obligation they could, for example, only fix security issues that affect most users and do not have a workaround.
> GPL 3 or AGPL do not help with this problem. it might stop a few big businesses, but again, it would stop them from using it
The big businesses are the one who should pay if they use it, right? What are they going to use instead? It would be cheaper to pay a licensing fee than to develop their own alternatives to things like OpenSSL or xz (the examples given that I was replaying to).
Edit: also, if you look up the thread I was replying to someone who specifically said the problem was with MIT style licences. At the very least GPL licences will prevent them using without contributing back their own fixes and, with GPL3, also has other positive effects.
> it might, now that they are already using i
but the problem is that they are mostly not doing so, right?
And I think this point is a poor as any made in the article.
Last I attempted to write some smaller ocaml project I used LLMs for support (but wrote myself). They generally weren’t excellent.
Honestly, types don’t help as much as people want them to. The LLM does best on popular languages, especially those whose use and feel is also mainstream.
That is to say, pick a niche language like Odin, and it may incorrectly start to over-apply Go’isms - knowledge from one language bleeds into how it approaches others.
But the US consumer is also quite self-centered. They will happily torch alliances and neighborly relations over perceived personal benefits. I doubt the US consumer would lift a finger for Taiwan. It will be a combination of weakness and indifference.
I just asked chatgpt today for research on whether any feature of a programming language which has been identified in studies to make it better suited for agentic coding. Or whether popularity in the data-set has been proven to matter or even cross-language learning (knowledge from one language benefitting the LLM in writing in another).
Aside from popularity, nothing I could dredge up could prove strong correlations with anything else - including types
reply