What is incredible here is that a UI developed in the 1980s is still attractive, and is arguably better looking and more usable than Windows 11 and Mac OS 12. What does that say about the last 40 years of usability and user experience research?
Sure, but I'd say the problem is it also left us with the idea that "because Mom's iPhone is so popular, everything, no matter how complex, should look and feel like that," leaving out e.g. a great deal of useful complexity.
I think complexity doesn’t have to be exposed by default on the UI. I’m not an apple fan, but I think ios finds a great balance between not having buttons for every single command (though behind menus it would be ok), yet still exposing it with some niche tricks you have to learn. For example touching the upper left corner’s clock will scroll up, which won’t be discovered by my mum but I can use it.
The thing I hate most about Apple products is these secret member handshakes that are completely non-discoverable to anyone outside of Apples precious alumni. I’ve been using Apple products for years and I still haven’t figured out hall of these tricks and nor can I recall them when I need them. In fact I’ve even got as far as accidentally triggering gestures and thought “that’s handy” but then been completely unable to figure out how to re-trigger that gesture.
I really don’t understand the hate for menus. Everything is named and categorised. Ok it doesn’t translate to a mobile phone all that well but I also don’t see why desktop UIs have to mimic smart phones and tablets.
For me, KDE gets it right. All the key features immediately available. Menus for advanced features. Hot keys for anything useful (with those hot keys names in menus). And terminal integration with all their core desktop applications (like file manager, text editor, etc) so you don’t have to compromise between using the GUI or CLI.
KDE also makes tiling windows easy without forcing people into a full on strict tiling WM. Something macOS is appalling at - even Windows does a better job than macOS.
I think we should differentiate between single-use programs and those that are more of an investment. Like, my torrent client should not be too clever as I will use it haphazardly with way too much time between starts, so anything I should memorize about it will likely be forgotten by the next time.
But on the other hand, things like photoshop, blender and even microsoft word and general uis absolutely fall into the investment category. You have no chance to discover by accident how to create a complex scene in blender for example — “pros” will always have new tricks for you. These are very complex programs where even your hierarchical menus would overflow with info — so now you have context-dependent right clicks bringing out menus not available otherwise. And I think it is a really great design, originating from the very first GUIs! But it only makes sense for us because we have already learned over the years/decades that right clicking is a cool trick oftentimes providing more functionality. It is a cross-program convention. While available on the menu bar, alt+tab is also not an instinctual thing to press, we had to learn that as well.
Now phones did have to reinvent many things and whether we like it or not, newer generations’ first instinct will be touch based, not mouse cursor based. I also see no evil in trying to create UXes that could potentially benefit from both touch and cursor-based usage. These new conventions has to be created though and hopefully one standardish way will emerge.
> I think we should differentiate between single-use programs and those that are more of an investment.
No you shouldn’t. What might be single use for Bob might be a daily driver for Bev.
The Ribbon bar is a great example of that. It’s targeting people who use applications daily but I only use Word or Excel sporadically and really miss the old menus and toolbars. But there’s going to be tools I use daily that others might only use sporadically.
If only there were a way to let Bob use what Bob likes and Bev use what Bev likes? Like giving people options? Like setting up a nice default, but also providing a path for individuals who want to change things in a meaningful way if they choose?
I'm sorry, but it's REALLY hard to not be snarky about this. Apple has absolutely suckered the world into thinking that "nice default interface" and "ability to change your interface" are somehow zero-sum, and there is no reason for this to be the case.
This deficient thinking has even wormed it's way into Linux, look at the current direction of GNOME. Sigh.
I have a problem with GNOME specifically because it changed courses so drastically; it was open and friendly and then by GNOME 3 decided not to be. Bit of a bait-and-switch in my opinion.
Project authors need to have the ability to own the project that author if there’s any hope of it growing and evolving with the times. We might not always agree with the leadership in a particular project but as like as it is open source that’s not a problem as anyone can then fork it and add their own vision to the fork. Which is precisely what happened with GNOME.
As an author of FOSS software myself, I’d quit in a heartbeat if someone told me I wasn’t allowed to make my own decisions about my own projects.
The ribbon bar is much more directed at discoverability for beginners, while the old design corresponds to features hidden behind menus under menus. So you being a sporadic user would (should) fare better with the Ribbon bar, it’s just probably that you remember the older version and change is painful.
It’s not the change that’s the issue. It’s the lack of context for what icons do without clicking them. Text offers that context.
But maybe the ribbon wasn’t the best example of my earlier point about the UI needs of sporadic users vs regular users. Or maybe it illustrates the point really we’ll but the context should be casual users vs power users rather than the frequency of usage?
You might be right. You might be wrong. I have no clue. And it doesn't matter a whit.
The point is - LET PEOPLE CHOOSE. Apple and Windows generally do not do this, Linux does. Or at least used to; this is why I'm actively hostile to things like GNOME getting more popular.
> The thing I hate most about Apple products is these secret member handshakes that are completely non-discoverable to anyone outside of Apples precious alumni. I’ve been using Apple products for years and I still haven’t figured out hall of these tricks and nor can I recall them when I need them. In fact I’ve even got as far as accidentally triggering gestures and thought “that’s handy” but then been completely unable to figure out how to re-trigger that gesture.
True, discoverability is an issue. Sailfish OS suffers from the same. I guess any gesture-based UI does, and on top of that iOS, Android, Windows, macOS... they all have features which the user is unaware of, and when they discover it its like 'why didn't you tell me, Ben?'. Programs do, too. Heck, you could say the same about programming languages. But in the end, it is you, the user's fault. Or is it? Perhaps that is part of the problem?
> I really don’t understand the hate for menus. Everything is named and categorised. Ok it doesn’t translate to a mobile phone all that well but I also don’t see why desktop UIs have to mimic smart phones and tablets.
Because designers instead of two distinctive UIs want a UI which is mostly the same for non-touch and capacitive touch. I saw this in the Maemo community as well. You had three types of apps (ignoring resisitive/capacitive focus, Gtk to Qt transition, CLI apps, web apps (as far as they existed) etc etc): normal/traditional desktop apps ("ugly ports"), touch-optimized apps, and stylus-optimized. That's a mess. And if you take for example iOS, that's where the majority of Apple applications are targeted for; not macOS. So they added a portability layer in 10.14 Mojave.
We may safely say stylus-based is not in use anymore. Back in the 90s it was used for resisitive based touch. Nowadays, if one uses a stylus, its on a capacitive touch screen, and it resembles a finger. So, ignoring web apps (agnostic due to CSS / responsive design), CLI apps (niche) we got two remaining: mouse/pointer based UI (traditional/normal) and capacitive touch-based UI. The one which became popular since iPhone is capacitive. It has taken over the world, so to say.
As a desktop user I use Rectangle on macOS, Power Toys on Windows, and on Linux I would like to use Sway with KDE in the future but right now I am sticking to Gnome. Gnome did spatial file manager right back in the days. Sometimes it was annoying, I think a tiling WM is a better solution nowadays, and KDE allows for that (on top of that Qt > Gtk).
Anyway, neither Windows nor Linux have an Apple Magic Trackpad 2 with gestures which work. macOS does. They're consistent. They work, both hardware-wise and software-wise. But it is not an iPhone with capacitive screen, and given the removal of TouchBar I doubt Apple will go towards such a path. You see, that the fingers cover part of the screen of an iPhone is a disadvantage of the device; not an advantage.
> True, discoverability is an issue. Sailfish OS suffers from the same. I guess any gesture-based UI does, and on top of that iOS, Android, Windows, macOS... they all have features which the user is unaware of, and when they discover it its like 'why didn't you tell me, Ben?'. Programs do, too. Heck, you could say the same about programming languages. But in the end, it is you, the user's fault. Or is it? Perhaps that is part of the problem?
This is about the (lack of) discoverability of UIs, including gesture-based, touch-based UIs. It also describes such lack of discoverability isn't a new phenomenon in software.
> Because designers instead of two distinctive UIs want a UI which is mostly the same for non-touch and capacitive touch. I saw this in the Maemo community as well. You had three types of apps (ignoring resisitive/capacitive focus, Gtk to Qt transition, CLI apps, web apps (as far as they existed) etc etc): normal/traditional desktop apps ("ugly ports"), touch-optimized apps, and stylus-optimized. That's a mess. And if you take for example iOS, that's where the majority of Apple applications are targeted for; not macOS. So they added a portability layer in 10.14 Mojave.
> We may safely say stylus-based is not in use anymore. Back in the 90s it was used for resisitive based touch. Nowadays, if one uses a stylus, its on a capacitive touch screen, and it resembles a finger. So, ignoring web apps (agnostic due to CSS / responsive design), CLI apps (niche) we got two remaining: mouse/pointer based UI (traditional/normal) and capacitive touch-based UI. The one which became popular since iPhone is capacitive. It has taken over the world, so to say.
This is about how we transitioned to a capacitive touch-based UI which has become the status quo, and how developers want to develop for both traditional pointer-based UI and touch-based UI (or how its made easier for them to do such).
> As a desktop user I use Rectangle on macOS, Power Toys on Windows, and on Linux I would like to use Sway with KDE in the future but right now I am sticking to Gnome. Gnome did spatial file manager right back in the days. Sometimes it was annoying, I think a tiling WM is a better solution nowadays, and KDE allows for that (on top of that Qt > Gtk).
> Anyway, neither Windows nor Linux have an Apple Magic Trackpad 2 with gestures which work. macOS does. They're consistent. They work, both hardware-wise and software-wise. But it is not an iPhone with capacitive screen, and given the removal of TouchBar I doubt Apple will go towards such a path. You see, that the fingers cover part of the screen of an iPhone is a disadvantage of the device; not an advantage.
This is my recent [anecdotal] experience regarding window management and GUI flow on pointer-based UI. Includes which tools I use and recommend.
You, in passing, nailed exactly what's wrong with Apple's way of doing UI.
"Niche tricks you have to learn?" That's ridiculous. It's perfectly fine to do a great default UI that most people like without any adjustment -- but it's preposterous that there's no "go in and change just about whatever you want how you want (even if that option is in the way back)."
This is why I've been hardcore Linux for quite some time.
I didn’t think this was niche, I’m pretty sure this has been an iPhone feature since day 1. I used to trigger it by mistake on the small screens.
It’s also not really a critical gesture, it’s a power user feature like keyboard shortcuts on a desktop system. It’s there if you need it but you can also grab the scroll bar and drag it to the top to do the same thing.
> This is why I've been hardcore Linux for quite some time.
Good thing Linux has no niche tricks that have to be learned to use it.
I kid, I love Linux, but this is a weird criticism to have of iOS which, interface wise, is far simpler than any Linux distro I’ve ever used.
Someone said it much better way up above but -- yes, Apple has "niche tricks that are hard to even know about, because Apple actively hides things from you and secretly changes things without telling you."
Linux is "Yep, this isn't easy, but it's generally well documented and talked about. It may take some work to get there, but it won't be mysterious and unexpected like Apple."
Could you sit in front of Blender, Microsoft Office, GIMP or Photoshop without previous knowledge of them and do anything at all with them? What about IDEs? Please see my other comment where I go into more detail, but I think sufficiently pro features can’t really be made discoverable and it’s okay to have only the base functionality exposed. And I don’t think linux changes anything over that other than you being able to use a different desktop env/window manager. I have had my fair share with tiling wms and whatnot, but on linux I now default to Gnome. It is opinionated but I think that’s the job of a good wm.
This is one of the things I really dislike about my iPhone. I've had it over a year, I've built apps for it professionally for close to a decade (cross platform, so I usually was the teams 'android' guy, but I still had to do iOS cause I knew it...) and I'm _still_ finding things almost every few weeks I had no idea existed. Like the ability to pick up multiple icons on the home screen.
I think comparing computers (as in the PC definition) and phones is exactly the issue.
I’ve nothing against iPhone (I’m even writing this answer from one) but it’s designed for consumption, which is ok, for a phone.
But I think that, step by step, we are loosing the good ol’ PC. And to me the good ol’ PC with its apparent complexity, is basically the last open and hackable (on multiple levels) technological tool left to the masses.
PCs main feature is that it is an infinite-purpose tool. You can’t be an infinite-purpose tool without some apparent (but logical) complexity.
Simplification of PCs UX is not even needed anymore since it’s being replaced by other devices for the people that just want to consume.
Anecdotally to this anecdote about iOS usability, my mom didn’t know about the control center or how to switch from wifi to cellular data.
iOS isn’t easy to use because Apple markets itself as special and smart, instead it’s easy to use because we live in an age where people use computers and can guess at most things.
I actually thought nextstep looked really ugly even back in the day. I really dislike the icon styles for some reason with so much grey. Looks really dated (and I can remember thinking that when it was out!).
I don't mean "out" as in first release date, I mean "out" as in it was currently being updated at the time. You're right on Win2.x and System6, but later versions of Win & Mac both looked better to me.
The very last version of NeXTSTEP was 4.x, and was released in 1996 near the end of NeXT's life. It was basically identical in looks to 3.x. Here are some screenshots of 3.x:
Windows was of course presently on Windows 95. Interestingly, both Windows 95 and OS/2 had largely copied NeXTSTEP design elements, including its famous "3D" look and feel, but even little things, like the popup corner menu, practically identical close and minimize (er, maximize) icons, etc., but with a much worse graphics subsystem, and terrible fonts and icons compared to NeXT.
MacOS was not quite on version 8 yet but let's round it up. It too had copied design elements from NeXTSTEP, and likewise had a much cruder graphics subsystem with terrible fonts and icons in comparison.
Beauty is in the eye of the beholder of course. Windows and MacOS had caught up somewhat by then. But I think it's pretty incredible to seriously say that Windows or MacOS were ever better looking (or functioning) than NeXTSTEP, at least up until NeXT was bought by Apple and NeXTSTEP ceased to exist as a product. Heck, neither 95 nor MacOS 8 even had live window dragging, something NeXTSTEP had in 1989.
I have a theory that NeXTStep is so grey, is because high resolution monitors with sharp text were only really affordable in monochrome. So the theming originally leant into that.
Yes, when I started using Linux in ‘99, several desktop UIs imitated this style. And it looked dated and ugly to me, compared to Windows 95, KDE and Gnome.
Of course, it's a matter of taste, but GNOME and KDE looked pretty bad back then. If I had the money, I'd definitely preferred NeXTSTEP. It's just so much cleaner and better thought-out. This is what it looked like in the early 90ies:
I don't know that I entirely agree. It's a bit plain Jane - reminds me of CDE in the 90s - but it's not unattractive. I can see it for a more focussed work environment, maybe, but I could say the same of a terminal running fullscreen in a modern OS too. Or just unplugging.
That's only for a certain degree of "use", one most users sadly don't (or can't) get past. Too much good from the past has been buried in the name of making it "easy" for people to do basic things; getting to the really useful things is becoming either a hoop-jumping exercise or no longer allowed.
That's like saying 100 years of research into music theory gave us autotune. If people could have made a capacitive touchscreen handheld device in 1980, I'm sure the interface they would have arrived at would be remarkably similar.
You know that the "better looking" part is totally personal, don't you? Personally, I never liked NeXTSTEP looks, not even 25 years ago when I saw some reimplementation under *nix (never touched a NeXT workstation unfortunately).
Indeed. I was around at the time this was considered beautiful. I didn't like it then, and I don't like it now. I guess not sharing the same taste or fanboyism is offensive. Good thing you weren't flagged. The night is young though
Edit: case in point haha. The squad is working overtime to protect the precious from evil non positive words
How dare you deny the Hamburger? And two paths moved through the forest, and our civilization took the Hamburger Menu path. And nothing was the same ever. What a thing the hamburger menu is. One can only divide the history of UI between the pre-hamburger and post-hamburger days.
Plus, I love how we went from 100Mhz to 4 gigahertz and waaayyyy more parallel issue, and yet I still routinely have delayed typing output in 2021 on almost every device and computing platform I use. At least several times a day I count: 4 billion, 8 billion, 12 billion 16 billion... and then my text appears and I can resume typing.
NeXT was great. The Magnetic drive, maybe not so much. It was the original borg cube!
> NeXT was great. The Magnetic drive, maybe not so much. It was the original borg cube!
There's some lost lore there: the "beachball" cursor that still lives in macOS was a pictogram of a spinning MO disc. It was a lot more obvious in Mac OS X 10.0 and 10.1 before it was redrawn into Aqua-style in Jagwire.
> What is incredible here is that a UI developed in the 1980s is still attractive, and is arguably better looking and more usable than Windows 11 and Mac OS 12.
I somewhat agree. To me, the utilitarian GUI options and desktops have always had some appeal: i've found the likes of LXDE and XFCE to be wonderfully usable and still have a soft spot for the Redmond theme, alongside boring looking pieces of software that just fit in really well with the overall theme and style of the OS.
Of course, that's a subjective point and opinions might differ: even in regards to *nix, some much rather would work with something like KDE Plasma, GNOME or something else entirely, and that's okay.
It would be interesting to have some insight into how much UX / UI goes into macOS and Windows releases, especially if things are tested with representative, non-technical users and then sent back for changes. My cynical take aligns with yours, for every clever small touch there seems to be a half-dozen awkward design choices to work around.
Not enough if the Dock is any measure. It’s been bad for over 20 years and in some ways still a downgrade from the NeXTSTEP Dock (and to be fair, in some ways an upgrade since it has a few tricks now the NeXTSTEP Dock didn’t).
Same for the Finder. There’s been ample opportunity to revisit the original compromises made, fix some of the behavioral and consistency issues and just not touch it ever again. That, hasn’t happened. To be fair, Spotlight and Quick Look are major feature upgrades over the initial Mac OS X Finder, but pretty much all of the criticisms from 20 years ago, well documented by John Siracusa at one point, stand. Also nobody wants .DS_Store files on their servers and it’s probably worth revisiting that decision.
Problem is turning them off kills your window positioning and view options which are pretty important to me. So I want them, but I don’t want them, but rather an alternative to what they do for me that doesn’t scatter files to every single disk or server I look at in the Finder.
A lot of UI development is resume-driven, or marketing-driven. Users buy hardware based on UI appearance, and regular changes make it feel fresh and new.
Those of us who use computers for more serious types of work generally prefer more consistency.
Guess there's no accounting for taste, it's clearly an ugly developer designed UI. All for maximum UX but prefer not to spend most of my day looking at something so aesthetically displeasing.
Cool nostalgic retro project though, looks similar to first Unix OS's CDE Desktop I used 20+ years ago.
That's exactly it. Taste, you either have it or you don't.
NeXTSTEP was good enough to pass the Steve Jobs taste test. Outside of Apple products, I double anything you consider "aesthetically pleasing" could say the same.
One productivity hack I discovered for myself was to set my background to a solid color I like and to make my windowing experience as "boring" as possible. Id otherwise sit and tweak my UI instead of working.
This sort of project is quite possibly the next step in that optimization for myself.
My favourite, and I kinda miss it, was a neat little hack/exploit of a new Windows feature in 95/NT4 + PowerToys. It gained a new option to stretch the wallpaper to fit the screen resolution.
If you set the wallpaper to a tiny 2x3 BMP -- literally just a rectangle of 6 differently-coloured pixels -- then in its efforts to scale that, it turned the background into a cool smooth set of blends, just a wash of gradients.
It was attractive but totally undistracting, and a ~6 byte wallpaper file took an extremely small amount of RAM. :-)
Eventually, if I remember correctly, it got "fixed" and the scaling algorithm recognised sharply differing pixel colours and scaled them into GIIIAAANNNT PIIIXEEELLLS. :-(
I have found my people! The solid black wallpaper reinforces in me the idea that computers and phones are tools, not accessories and it has been revolutionary. I now try to use more and more things in their default state. Vim instead of NeoVim, and an empty vimrc. The shitty fonts that ship with the distros, rather than Source Sans 3. That hopelessly ugly yellow Matlab Current-Section highlight colour. I love them all!
It was staggerging to me just how much easier it was to stop, breathe, take a moment to change my mind about the fact that some config needs changing, and instead just say "it's good enough".
#444 since 1997, more or less: For some years I used a 1 pixel high image with a single black pixel tiled vertically to delineate the area where I put Window Maker's* dock.
*: Window Maker is a window manager under X11 with a NeXTStep-look-and-feel.
On Linux a window manager might be another option in that direction. I use i3 (with a solid #1F4437 background :) and find it's the best way to make your desktop invisible.
I get the Bing Wallpaper of the day on my desktop (be it Windows, macOS, or Gnome). I love it because it is consistent on my devices (except (work)phone and Citrix). Its a surprise every day, usually positive. My oldest likes it, too.
Interesting, I find that I spend a day or two getting everything 'right' for me, which tends to end up in this exact state where it's both boring and pleasant to look at. I also use a solid black background.
Yes. I have a similar mindset and reasoning for using Mate desktop. It's very simple, I'm spared the complexity of kde and the eye candy / novel paradigm of gnome.
> Workspace is NOT WindowMaker... Workspace is written from scratch
Kinda, but not exactly and for some reason the author has always been vague when it comes to that.
Nextspace's Workspace has been written "from scratch" in the sense that it is a new (sub)project, however it is meant to be the full desktop experience instead of just a window manager. The window management part however is based on a fork of Window Maker, you can find the relevant files (with their copyrights, etc) under Applications/Workspace/WM. The build system has changed to Nextspace's (which seems to be just GNU make) and some glue code with the rest of the system is made.
I don't know, but I think GNUstep deserves more appreciation and use.
Modern hardware is drastically more powerful than what NeXT had to work with, yet it often bogs down just trying to implement a usable and responsive GUI.
Sadly my work laptop is on Ubuntu (macOS at home)
GNUstep inspired me to try Chicago 95, so I will probably go with that later on... Nextspace looks amazing tho'
Thanks for the mention of Chicago95! I hadn't heard of it before, looks like just what I wanted to try. The various early windows themes for things like icewm left me wanting a bit more "authenticity".
But isn't it painfully wrong on the title bar size/font, the highlight around icons, etc? If this is the better one, I can't imagine the alternatives...
The titlebar size is fine, however the title is a pixel off vertically as is the icon too. The biggest issue i can see in the screenshot is the menubar spacing being too large and the scrollbar's top/left bevel lines using the "button shadow" instead of "button" color. In addition the bevel lines use the wrong overlap in the corner. For some reason this is only the case for the first button (up, left) and the second button (bottom, right) looks correct. The icon views (small and large) in lists also use too much spacing around the contents as well as between the icons and the icon captions in the desktop also use too much padding. Finally the window frame seems to also have too much padding between the bevel and the actual window contents (i.e. in the file manager views the outer bevel lines for the tree and icons should align with the titlebar, but here they are pushed a couple of pixels inside). Ah yeah, also the shortcut icon overlay seems too scaled down and blurry for some reason.
These are what i noticed from the screenshot. They're fixable for the most part though, i had played around with Chigago95 at the past to make it a bit more faithful. The biggest effort was modifying Gtk's theme engine to remove the outline from the menu bar :-P. And TBH i think that for pixel perfect accuracy you'd need to modify some code, which goes a bit outside the scope of a theme.
I am in no way nostalgic for 1990s OSes -- I remember the pain too vividly -- but
I think this look would be like a hangnail or a toothache: just wrong enough to be constantly irritating.
Now, 1980s OSes, when stuff was still a challenge, now that was fun. :-D
"Workspace is written from scratch. Some WindowMaker code is a part of Workspace (as well as configuration defaults) to provide window management functions. The code is tightly coupled with Workspace to provide seamless intergation. Configurable parameters of the integrated WindowMaker are spread across Workspace's Preferences and Preferences application."
If the answer to all 3 of those is "yes" then one should be able to port Solvespace to this UI fairly easy (it's one source file per platform). You know, if you want a CAD program...
Seriously, this reminds me of the worst of CDE and Motif, all not-quite-finished looking.
I ran a NextStep machine for a while (lipo is the best command name ever, IM(NS)HO) and loved it, and it looked nothing like this weird semi flat pseudo X11 ugliness.
(Only reason I switched away was that the rest of my co was Windows and loved themselves their Office docs and there was no equivalent for OpenStep.)
It kinda does - all the isometric drop shadows and whatnot. It's a lot cleaner (none of the weird beveled stuff) but all the sharp edged UI from that era look pretty much the same to me.
WindowMaker is a window manager. NextSpace is a full desktop with guidelines, specific behavior and utilities (file manager, terminal, inspectors, ...).
Man, the build instructions.. are mess. It depending on specific distro, is such... a turn off..
Too bad, because otherwise its amazing project, and I wish them the best...
Pain.
Here’s something I’ve been pondering: how did the first window managers visually fail when a program hung or had a fault?
With older Windows version, you got pretty GDI “trails” as you dragged a window around. Now you get blanked out elements.
For some reason, I thought that something like Plan 9 would have been flawlessly written (like their CLI counterparts) and that it would be immune to visual issues like that.
GDI "trails" are due to specific of how GDI drawing worked before WinNT6 and DWM.
Specifically, applications were issuing drawing commands which were executed directly onto the framebuffer - there was no "draw into rectangle then another process will compose it", because GDI was written for systems that simply didn't have enough memory for that.
Essentially the process looked like this:
1. Windows GUI system figures that a specific window (this could be everything from full application window to individual control) became visible
2. WM_PAINT (iirc) message is sent to the application, to specific window
3. the handler for WM_PAINT of the specific window is triggered (there might have been support for buffering pre-existing draws, too), and executed GDI calls to draw the component (canonically using a state pointer to associated data block with everything necessary for repaint)
4. The GDI system does the drawing directly on the graphic card (either fully in software, or through accelerated calls exposed by driver)
5. Next window that needs painting gets called etc.
This meant that when the applications hanged, or simply didn't have time to process drawing requests fast enough, you ended up with "trails" due to remaining old data in framebuffer.
With WinNT6+ and DWM, GDI applications each draw into their own buffer which is then composited by DWM. This means that at worst they break their own windows' contents, but not the global framebuffer.
What's particularly helpful in the case of WindowMaker (another NeXTSTEP environment, mentioned in TFA), is that the window manager itself can simply be restarted.
This is / was the case for several of the early-generation window managers, including twm, fvwm, and probably a few of the *Step window managers.
I suspect you might be able to pull this off with GNOME or KDE though I've not tried this in some time.
When restarting WindowMaker, the "naked" windows are briefly painted (without decorations) to the display, then decorations (frame, titlebar) reappear and windows resume their status (open/closed) and remain near their positions (the decoration toggle usually means they shift down and rightwards).
I have a color turbo. Wonderful. Nearly useless, but wonderful. I would like a Cube so that I can relive the wonder of its release at a Seybold tradeshow way back when I was early in my career.
At one point in the early 00’s I had twelve Cubes that I picked up for free from a company that was getting rid of them. I was flush with geek treasure, and wanted to generously share my hoard with friends. By the time I realized what I’d done, I’d give them all away. D’oh! But at least I made twelve people very happy, and I still have a big stack of Nextstep install CDs.
Wow, that is amazing. You belong in the geek pantheon for that. Two of the defining moments for young me were seeing the NeXT Cube in action, and seeing the original introduction of Adobe Acrobat. Seems mundane now, but grasping the potential of these things was profound.
No worries. Some people however are hoarding rare software - software that is very likely to cease to exist once their collections fail. But refuse to share or make backups.
It would actually be nice if the developers of these old applications would upload their source somewhere so that enthusiasts can build on them.
GNUSTep is an implementation of the Openstep spec, so there's a good chance that that old software could be ported to Linux (or even OSX) without too much effort.
Not really. Although there are of course similarities, NeXTSTEP as a graphical interface was implemented from the ground up by NeXT (Jobs' new computer company after he left Apple). The GUI of almost all other commercial UNIX operating systems was developed with the Motif toolkit and there was a whole desktop environment - Common Desktop Environment - which was popular for some time. Motif wasn't free though, so new GUI toolkits were being developed, most of them free, like GTK (initially for building the GIMP UI), and desktop environments were being created on top of them like GNOME. Because they were free and feature rich, Motif and CDE started declining and that's why you don't see them anymore as GUIs on UNIX based OSes.
Then Apple bought NeXT and redesigned NeXTSTEP's UI and called it MacOS X.
It's interesting and surprising to me that so many like the look of this. The fonts look grainy. The widgets are blocky and crude. At least, that's my take on the aesthetic. To each his own, I guess.
It's less that it looks nice and rather that there isn't anything surprising. No scrollbars that only show up when you mouse over them, buttons without borders, horizontal lists that aren't obviously a scrollable list... the list goes on. :)
In the PPC era I once installed X11 on 10.3 Panther and compiled KDE. I then had a MacOS X running a KDE shell and various X11 applications, while maintaining compatibility with OSX apps.
I'm wondering if something like this would still be possible on macOS 12 Monterey and using Nextspace as a shell?
a popular topic among the hn readers that comes up once in a while. i guess it starts with one and then people submit interesting links they found from that discussion
because window maker is not, and never was, anywhere near the look and (most importantly) FEEL of a real NeXTstep GUI. i have used both for a few years. window maker is a nice windowmanager for X11 but everything else is still standard X11 fare from style of widgets to the behavior. in particular, this matters for how X11 apps interact with the window maker dock.
an actual NeXTstep workspace and dock is quite something different.
NEXTSPACE is interesting specifically because it is not just window maker with some apps that emulate the look.
spending the effort to learn to have a keyboard driven workflow for common things as been tremendous for me. it effortlessly allows me to recreate my various workspaces (chat, music, email, browsers, work specific things, terminals, production, etc).
For coding I usually always dragged windows to the same setup, w/ i3, it's easy to automate that and control it all via the keyboard.
So you're not great at the whole reading-comprehension, are you? From, as they say, TFA
>For now, I want a fast, accurate, and stable version for RedHat-based Linux distributions. Currently these are CentOS 7, CentOS Stream and Fedora.
Those are the systems this thing runs on. The parent post I replied to wished they could run this on a raspi. I linked them to how to get CentOS running on a raspi, which would make the damned thing a supported configuration.
I have 3 persistent programs that tile and the rest float - when I had swaywm set to 'tile', it became such a burden to write 'for_window' clauses for every new transitory program.
If I want it to tile, it's easy enough to $mod-Space it.
On my daily driver, firefox & mythtv have their own workspaces. On another workspace, emacs gets a half-screen and 2 terminals get a quarter screen. I set them up at session start and that's 95% of what I need. After that, everything gets a floating screen - which is what many of the apps expect and need.
I also have some scripts that automatically summon floaters from the scatchpad and fit them on the screen (without changing their size, so not tiling).
Surprisingly many of them - not for me, but godspeed to them, configurability and freedom of X11 environment was big part of what made linux great for me.
