Hello together, longer period of time we have not replied back but now it is time for another news as times are really rough for everyone and also us in special. First and foremost: Hyperbola is not inactive and the development is on-going for both, our Hyperbola GNU/Linux-libre and HyperbolaBSD also.
But as machine-learning, LLM and in common named "AI" is enforced throughout many parts, software and also free, libre projects we needed to enumerate, wait, lookout for alternatives and more even then to decide. So this includes also that we now announce the complete removal of webkitgtk and all included based applications. The reasoning is already given: Webkit has an permissive AI-policy (1, 2, 3) granting AI-agents rights to change program-code and also uses AI-driven code-reviews (1, 2). This contradicts our substantial position as our project is from humans for humans, not from humans using the machine to review and create code in questionable ways. Also we follow the directive of minimalism: Lesser code, better review and audit-options included. To include much more code and later then being in some need to use external tools is the worst result of all as this makes unknown outcomes for dependending on more than only questionable and clear unethical services. Webkit itself contraditcs therefore our own stance. Yes, this comes out of a high cost as the so-called "modern web" is depending heavily on JavaScript. But we ask therefore back: Is this is a healthy course as we all have seen where this may lead up to? Breaching software, leaking data of users in high numbers and in the end risking even lifes of people in many ways.
Do we overestimate that? No, not even near and we cannot stop to repeat how important it is to stay in control of ...
While we have written before in a dedicated news-entry here nothing has clearly changed therefore on our stance towards rejecting generative, large language algorithms and neuronal networking.
As before we can only warn about the consequences, underline that no concrete model works without data not ever being enumerated on ethical and reasonable moral grounds. First and foremost you may ask: What does this mean in daily basis? Let's have a look on AI- and LLM-scraper or better also "crawler" bot-services (Link for more information).
Yes, those do exactly exist and those also do a tremendous harm as using doubtful ...
With the upcoming new version of Hyperbola GNU/Linux-libre we have removed, updated and renamed many common known packages, applications and libraries. Those updates will need manual intervention of the users, especially when it comes to the changed naming conventions. We have not included a direct update-path into the packages out of the reason, that we clearly want the users to take on responsibility also.
No worries as the way to remove and readd the packages is so far uncomplicated. First of all the changed naming-scheme with some examples to follow:
audacious -> audacious-legacyaudacious-plugins -> audacious-plugins-legacygimp -> gimp-legacy ...
After 7 years we are giving Iceweasel-UXP its final callout for a "Goodbye". Within our testing-repositories the package is already removed and so it will also be no longer part in the next upcoming version 0.4.5. We are very thankful for this fine webbrowser created as hard fork from original Basilisk and done from our founding member Luke (Gaming4JC). Time goes by nevertheless and we have no further maintainer for Iceweasel-UXP, so we are sending it with many thankful wishes to the deserved rest.
Thank you for being a companion on the web for such a long period ...
The year 2024 is running towards its ending and Hyperbola as project has made steady progress to its next release and also forming up our own BSD-descendant operating-system HyperbolaBSD.
Nevertheless we see the need to make a statement also about issues we all should be aware off: Free, libre software needs clear principles and values. Integration of problematic and non-free defined services for examples is no point for any discussion. Also the on-going debate about "machine learning" is not really promising in its current direction.
The newer the hardware itself, the more restrictive it is and the lesser it is ...
