Announcement

Collapse
No announcement yet.

Bcachefs Receives Funding By NLNet With NGI Zero Grant

Collapse
X
 
  • Filter
  • Time
  • Show
Clear All
new posts

  • rrveex
    replied
    Originally posted by jacob View Post

    This has also been proven to be a wrong approach. "Show windows, help the user organize them on screen" is what gave rise to the likes of TWM, 38476 forks of FVWM and the wayland-compositor-of-the-week. Unsurprisingly, no-one finds those to be a better designed UI than MacOS or even Windows. There is A LOT more to a UI and a good desktop than "show windows". I believe that the two most important features are 1) integration and 2) developer resources. Re 1., the requirement on a good UI is not just to run apps in windows, it's to make the apps work well together. Drag&drop, cut&paste (and not just for ascii text), settings management in the form of a unified event-based API rather than a bazillion of text config files with arbitrary syntax, accessibility features, plug&play that gracefully handles all the weird corner cases that happen 1% of the time but actually take 99% of the development effort: this is all what makes a UI people actually will use, as opposed to one that is fun to tinker with for people who don't have money to make.

    Re 2, a UI environment must make it trivial for developers to do the expected thing. A developer shouldn't have to write 1000s of LoC to make sure that the application works with languages that write right-to-left, or implement their very own piecemeal broken hack to support a single-instance app model. In many ways this is the Linux desktop's biggest failure. In the X11 window manager era, developing anything GUI-based was a total chore (even with the likes of Motif) because it was basically impossible to make the application behave sensibly at all without diving into the ICCCM shiteshow, and then it didn't really work anyway. Today we have an incomplete GNOME, an incomplete KDE that does things differently just because, and a developer culture that believes that the right way to develop an application is to reduce it to the lowest common denominator, which in practice often means discarding GUI altogether.
    Well, we obviously disagree on many aspects I won't get into all, it doesn't really matter, different people like different things, like I wrote before.

    "settings management in the form of a unified event-based API"

    Why t.f. would I want a *global* place in which to set editor tabstop=4? What if I want 4 in an editor and 2 in the other? How would a global place accomodate the miriad of settings one app has and no other (like vim, emacs, etc)? Still in the registry? And that specific place would be easier to find than ~/.config/app?

    Ruled by an over-complicated event API? Like we don't have enough to worry about, now we'd have to learn yet another cryptic useless thing? (like pam and pkexec)

    People don't even like the same language for config files; they keep inventing more easy to read and use languages, like json -> yaml -> toml. And this is a *good* thing, otherwise we'd be all still plagued by xml. Or binary - let the API take care of the format, what could possibly go wrong? After all, this new API would be perfect and bug-free (like systemd). If some program doesn't behave like you expect, you can always open a ticket, so you can be told that it's your fault, you didn't use the API properly and can go f.y.

    And so on. One of the nice things with free software is they don't have global management to impose *the one and only best way for all*.

    "Linux desktop's biggest failure" - there's no failure to talk about. Linux desktop is fine. The "great adoption" can't ever happen because of marketing/lobby/whatever. M$ & al will always push through other, much more effective, ways than quality. You have M$ in schools and institutions because 1. incompetence and 2. corruption. You have it in corporations because the decisions are mostly made by management knowing shit about software quality and being lobbied by M$. This is already the critical mass the many computer-illiterate will adhere to, just for safety (when something doesn't work, I'll just ask someone, "everyone" has Windows). And it's fine as it is. Wtf do I care that most people drink black-dyed sugar-water, aka cocacola? Let them, as long as they don't try to force it down my throat too.

    While I still can: tabstop=3.

    Edit: one more thing. Old people probably won't switch to Linux, because it's difficult to change when you're old. Young people could be attracted if it were cool, because this is what most young people care about. So if you care about "linux desktop failure", you should be glad about the many "cool" (and mostly redundant, but each "cool" in it's own way) wayland compositors; your "global event API" won't attract anyone
    Last edited by rrveex; 23 February 2024, 08:43 PM.

    Leave a comment:


  • jacob
    replied
    Originally posted by rrveex View Post

    I agree with Gnome and KDE having issues, like Gnome having no systray and KDE full of bugs in rolling distros (possibly ok in LTS distros, don't know).

    Now, what's a "good UI" (speaking of desktop)? Gnome keeps reinventing things. Does this make it better? Like "let's replace systray with ... " (I don't know, I still don't get it how you can replace system-tray with notifications... like, instead of an "on" icon, you constantly get reminders "app is still running"?). KDE tries to be as flashy as possible at the cost of stability - is this good or bad?

    I'd say "let the users decide what they like". Which makes the "plethora of me-too desktops and yet-another tiling compositor" a *good* thing. I can pick whatever I personally like best. I'd rather have 5 "me-too desktops" to chose from instead of letting almighty Gnome decide what changes to force on me.

    When a DE goes further than just "show windows, help the user organize them on screen", the extra stuff is nice for some and unusable for others. Gnome had a pretty good formula (except the "registry" settings), then they started "making it a better UI" and people fled to one of the many gnome2 clones.
    This has also been proven to be a wrong approach. "Show windows, help the user organize them on screen" is what gave rise to the likes of TWM, 38476 forks of FVWM and the wayland-compositor-of-the-week. Unsurprisingly, no-one finds those to be a better designed UI than MacOS or even Windows. There is A LOT more to a UI and a good desktop than "show windows". I believe that the two most important features are 1) integration and 2) developer resources. Re 1., the requirement on a good UI is not just to run apps in windows, it's to make the apps work well together. Drag&drop, cut&paste (and not just for ascii text), settings management in the form of a unified event-based API rather than a bazillion of text config files with arbitrary syntax, accessibility features, plug&play that gracefully handles all the weird corner cases that happen 1% of the time but actually take 99% of the development effort: this is all what makes a UI people actually will use, as opposed to one that is fun to tinker with for people who don't have money to make.

    Re 2, a UI environment must make it trivial for developers to do the expected thing. A developer shouldn't have to write 1000s of LoC to make sure that the application works with languages that write right-to-left, or implement their very own piecemeal broken hack to support a single-instance app model. In many ways this is the Linux desktop's biggest failure. In the X11 window manager era, developing anything GUI-based was a total chore (even with the likes of Motif) because it was basically impossible to make the application behave sensibly at all without diving into the ICCCM shiteshow, and then it didn't really work anyway. Today we have an incomplete GNOME, an incomplete KDE that does things differently just because, and a developer culture that believes that the right way to develop an application is to reduce it to the lowest common denominator, which in practice often means discarding GUI altogether.

    Leave a comment:


  • Weasel
    replied
    Originally posted by varikonniemi View Post
    I could not envision a better recipient for funding. Contrast to the recently announced 1 million for gnome? What will it materialize as?
    More features removed and maybe 2-3 complete UI overhauls/redesigns over the period of a year.

    Don't forget each UI redesign is "analyzed deeply and as perfected as possible by our specialized UI experts!". That's why we have so many redesigns in the first place. Perfection is very changing!

    Especially when you see them move elements from left to right instead. Did you know humans suddenly change their direction preferences overnight?

    Listen to the UI experts, they know what they're doing!

    (full sarcasm post btw, they're actually cringe useless wastes trying to justify their job to not appear so)

    Leave a comment:


  • yump
    replied
    Originally posted by varikonniemi View Post

    My point was more towards contrasting how frontend bs gets funding but the meat of the issue, what makes Linux the premiere, is most often forgotten in these kind of outside fundings.

    Sure, you want to fund Linux desktop. Best way to do that it by funding Linux, and let the desktop organically build on top of a incrementally better kernel you funded. Instead of spending 1M on UI you could spend 1M on kernel, and have 10M worth of enthusiastic volunteer UI developers come in because your kernel became next level.
    What focusing on the kernel to the exclusion of Actual Desktop Infrastructure gets you is a lot of really wonderful-for-desktop-in-theory kernel features that are some combination of 1) disabled by default, 2) tuned very conservatively by default, 3) are only useful in combination with a userspace control plane that only Android and hyperscalers have, and 4) disabled by default or mistuned on x86. For example...
    • Transparent hugepages
    • zswap (recently improving thanks to Facebook)
    • zram writeback
    • schedutil cpufreq governor
    • utilclamp (like Feral's gamemode, but per-process and able to work in reverse to limit background tasks to efficient CPU frequencies/cores)
    • energy-aware scheduling
    • PSI + cgroup memory controls
    • TEO cpuidle governor
    The feature that's in the best state AFAIK is cgroupsv2, which has a lot of support from systemd and Gnome/KDE for cgrouping individual services and apps.

    Leave a comment:


  • skeevy420
    replied
    Originally posted by fallingcats View Post

    Don't pretend windows has any functioning theming to speak if. Their dark mode barely works with apps. KDE at least manages to theme GTK apps, even if shadows look a bit off or something.

    Well, I never brought up Windows, at least in this thread in that kind of manner, but it does have accent theming; static and dynamic. It's not much, but it's something. Regardless of the DE, Linux themes better than Windows....Out of the Box. There are a shitload of ways to theme Windows when you go 3rd party; free or paid. UltraUXThemePatcher, all sorts of panel and start menu customization tools like Cairo Shell, and more.

    Because I've had this discussion about Windows themes with two different people this week: It ain't my fault that all y'all Linux people cain't Google the words "windows theming" and follow the rabbit hole. It ain't my fault y'all got as far as the System Settings menu and gave up. I have SMH induced whiplash.
    Last edited by skeevy420; 21 February 2024, 10:23 AM.

    Leave a comment:


  • LinAdmin
    replied
    Originally posted by fallingcats View Post

    Is it really that hard to look for yourself?

    https://bcachefs.org/Roadmap/
    Thx for this link, I had overlooked it ...

    Leave a comment:


  • ehansin
    replied
    Originally posted by andyprough View Post
    Pretty cool to see what that old power plant has been turned into with all the rooftop gardens and shops and restaurants and apartments and everything.
    I'll be honest, I didn't know the name of the thing off the top of my head, but looked it up. I didn't realize they redeveloped it, but after reading this I looked it up again and dug in a bit. I've never been to England so have of course never seen in person. But if I make it some day, I'll have to have a look!

    Leave a comment:


  • rrveex
    replied
    Originally posted by jacob View Post

    That approach has failed. The outcome after 30 years of that is that we have GNOME which has a number of issues, KDE with another number of issues, and a varying plethora of me-too desktops and yet-another tiling compositors that bring little value to the big picture. The kernel's funding is there, it's plentiful and it's important, but to get a good UI, you need to directly fund the UI, not hope that it will somehow magically appear out of thin air.
    I agree with Gnome and KDE having issues, like Gnome having no systray and KDE full of bugs in rolling distros (possibly ok in LTS distros, don't know).

    Now, what's a "good UI" (speaking of desktop)? Gnome keeps reinventing things. Does this make it better? Like "let's replace systray with ... " (I don't know, I still don't get it how you can replace system-tray with notifications... like, instead of an "on" icon, you constantly get reminders "app is still running"?). KDE tries to be as flashy as possible at the cost of stability - is this good or bad?

    I'd say "let the users decide what they like". Which makes the "plethora of me-too desktops and yet-another tiling compositor" a *good* thing. I can pick whatever I personally like best. I'd rather have 5 "me-too desktops" to chose from instead of letting almighty Gnome decide what changes to force on me.

    When a DE goes further than just "show windows, help the user organize them on screen", the extra stuff is nice for some and unusable for others. Gnome had a pretty good formula (except the "registry" settings), then they started "making it a better UI" and people fled to one of the many gnome2 clones.

    Leave a comment:


  • varikonniemi
    replied
    Originally posted by jacob View Post

    That approach has failed. The outcome after 30 years of that is that we have GNOME which has a number of issues, KDE with another number of issues, and a varying plethora of me-too desktops and yet-another tiling compositors that bring little value to the big picture. The kernel's funding is there, it's plentiful and it's important, but to get a good UI, you need to directly fund the UI, not hope that it will somehow magically appear out of thin air.

    Plasma 6 is lightyears ahead of something like windows. Only mac might still be more consistent in UI look. And both of them also have issues.

    So what are we really comparing to when saying failed? IMO almost every Linux desktop is better than the proprietary competition because they lack customization, so they failed from the get-go. Imagine how much easier and more bug free gnome 3 would have been if they never introduced a extension interface.

    Leave a comment:


  • fallingcats
    replied
    Originally posted by LinAdmin View Post

    Any URL giving details available?
    Is it really that hard to look for yourself?

    Leave a comment:

Working...
X