Nice post. Relatedly, see also malus.sh and this talk by the people that made it (both of which I posted in this lemmy community here).

A couple of minor corrections to your text:

Blanchard’s account is that he never looked at the existing source code directly.

Blanchard doesn’t say that he never looked at the existing code; on the contrary, he has been the maintainer (and primary contributor) to it for over a decade so he is probably the person who is most familiar with the pre-Claude version’s implementation details. Rather, he says that he didn’t prompt Claude with the source code while reimplementing it. iirc he does not acknowledge that it is extremely likely that multiple prior versions of it were included in Claude’s training corpus (which is non-public, so this can only be conclusively verified easily by Anthropic).

The GPL’s conditions are triggered only by distribution. If you distribute modified code, or offer it as a networked service, you must make the source available under the same terms.

The GPL does not require you to offer GPL-licensed source code when using the program to provide a network service; because it is solely a copyright license, the GPL’s obligations are only triggered by distribution. (It’s the AGPL which goes beyond copyright and imposes these obligations on people running a program as a network service…)

Nice, thanks.

It would certainly be nice to be able to pre-download language pair models without selecting to and from and then actually initiating a translation using the model i don’t have yet.

re: getting uBlock externally, i also see the attraction of that approach but unfortunately Debian’s package was last updated in October (from 1.62 to 1.67) while AMO has a release from January (1.69) :/

imo it would be better to bundle UBO and ship its updates along with browser updates.

are there plans to distribute Konform via flathub?

Full-page machine translations are disabled

Firefox translations are done offline (after downloading the model for a langauge pair).

Does anybody know why Konform decided to disable this very useful feature?

It’s a library for detecting which character encoding a string is encoded with.

Here are the docs for the vibe-coded rewrite, and here is the version before it.

The new vibe-coded version also adds language detection; it isn’t clear to me why the current version of the readme shows it classifying the string "It’s a lovely day — let’s grab coffee." as Spanish with 99% confidence, without any comment in the docs about that being a misclassification, but I guess that if the LLM-authored program says it is then that must be one of those phrases that looks the same in Spanish as in English 👀

Why do you think Proton stores the association between accounts and payment identity?

Many privacy-oriented companies actually accept credit card payments and simply don’t store that information.

article in case you can’t read it: https://lemmy.ml/post/44086795

that link only has two paragraphs of the article; there are 8 more in the full article here on archive.org

is it time for a Windows edition of the classic Jamiroquai sound meme?

could Red Hat eventually take control of the project

Fedora started in 2002 and merged with “Red Hat Linux” in 2003.

Red Hat, Inc has had full control of it ever since then.

It is a “community project” inasmuch as there are Fedora developers who are volunteers (and some who are paid by companies other than Red Hat), and the Fedora Council includes people who are not employed by Red Hat - but the Project Leader is always a Red Hat employee, and if the Council ever has an irreconcilable difference with Red Hat then Red Hat can simply ignore and/or dismiss them.

Red Hat owns all Fedora-related trademarks, and the Fedora Project is not an independent legal entity: it is a part of Red Hat.

If Fedora developers don’t like Red Hat’s decisions regarding the project, they can fork it but they’d need to change the name and find some other sources funding.

Also, icymi, Red Hat became a subsidiary of IBM in 2019.

im not sure how easy it is to get LLM’s to output near-verbatim

the more relevant question is how easy it is to be confident that they aren’t. (and the answer is: you can’t)

i dont want “using AI” to be used to undermine my efforts

😭

what you see is the result of the typical software development process

note that many free software projects have had to establish rules explicitly prohibiting LLM contributions precisely to avoid the issue of plagiarism, as well as the related issue of copyright infringement. (and of course, code quality is another reason to…)

you’re the first to give any pushback on it

alias thanks="You are most welcomed"

you need to use the echo command if you want to echo something:

alias thanks="echo 'You are most welcomed'"

(the inner single-quotes are not strictly required in this case, but recommended nonetheless)

https://gitlab.com/edneville/please (a sudo alternative)

given the fact that LLMs’ outputs can include verbatim or near-verbatim excerpts of their training data, how do you justify placing an open source license (or more generally, any copyright claim of your own) on source code you’ve “authored” using one of these automatic plagiarism machines?

I am using AI to do this so its important to be clear that im not asking you to audit or review my vibecoded project. While i have made attempts to create things like audits and formal-proof verication, i know better than to trust AI when it tells me its working correctly.

Sharing it now if there is feedback about the implementation. Anything i may be over looking? Feel free to reach out for clarity on any details.

Did you write this post using an LLM too? You say that you are “not asking you to audit or review my vibecoded project” and then in the very next paragraph request “feedback about the implementation”.

Here is my feedback: stop trying to vibe-code privacy software.

i know better than to trust AI when it tells me its working correctly

if you know better, why are you continuing to do it, and to post your outputs here?

what’s the goal?!

maybe it would be better to say that it is stochastically accurate?

also btw icymi, this is a post about LLMs

!memes@lemmy.ml mod here - rather than downvote, i deleted it.

imo “eliminates 99.9% of [demonym]” isn’t funny or ok for any value of [demonym], even when you’re making a germs/germans pun.