Comment by meowkit
4 years ago
I work on the core OS.
This is the answer. Now it could be done to maintain backwards compat and have everything completely up to date, but the juice is not worth the squeeze and there are more important things to focus on (such as making the OS work/build in such a way that there won’t be inconsistencies in the future).
Asking the obvious question: why aren't the UIs in newer Windows versions not "just" different themes over Win32 widgets? Instead the UI theme seems to be tied to UI frameworks which seem to be created and discarded annually at Microsoft, so each time some genius decides to overhaul the UI, a completely new sediment layer is created instead of letting some UI designers tinker with a theme editor.
I don't get it. Functional UI changes should be different from visual fluff, yet Windows seems to the only OS where the two aren't separated.
This was historically the case, it's why despite the article calling out winver, folder properties, the run dialog as being from the win9x -> win2k period, they didn't look remotely as out of place in win7 as win10.
Of course there are limits. iirc Windows 7 RTM had an obviously 3.1 derived file picker make it into the add fonts dialog until it was replaced in SP1 if my memory has the timing right. Despite using win32 widgets which had the correct theming, that didn't look native at all
FWIW: I consider fixes like the improved environment variable editor in the "old" system control panel in Win10 much more important than the entire new settings panel mess. The same type of improvements would make a lot more sense for the rest of the builtin UIs instead of creating entirely new "construction sites" all the time, which will be left incomplete and outdated in a few years time as the current people are replaced with new people.
There is no such thing as canonical “win32 widgets.” There have always been several different UI libraries even for win32. The original win32 styling you get with CreateWindow doesn’t support modern font rendering, high DPI, doesn’t work well with touch, is only software rendering, doesn’t support latest accessibility features, etc.
The newer stuff sits on top of DirectX.
The latest UI framework (XAML/WinUI) sits on top of a better rendering architecture (compositor) that itself wraps DirectX tech.
There's no reason why the old Win32 window and widget foundation couldn't have been lifted on top of a modern rendering architecture though, there was certainly enough time to achieve this. This would also reduce the need for creating entirely new UI frameworks every year (or even better, higher level UI frameworks could be built on top of a modernized but backward compatible Win32 widget foundation instead of doing all their own rendering directly on top of DirectX).
(TL;DR: the Windows UI mess looks to me like an organizational problem, not a technological problem. From the outside it looks like as the old guard is leaving, the new people coming in discard all the old code and instead prefer to start completely new projects, which of course is destined to fail, because the new people are repeating the same errors which had been identified and fixed decades ago).
3 replies →
> The latest UI framework (XAML/WinUI) sits on top of a better rendering architecture (compositor) that itself wraps DirectX tech.
WinUI/UWP still uses CreateWindowEx and good old WndProc for event loop message handling at the top level on Windows 10.
I guess one challenge with this is that old school Win32 doesn't really support anything resembling automatic / responsive layout. All controls are of a fixed size in a fixed pixel location. This means you can't for example change the default font, add margin or padding or do any other significant changes without essentially breaking all existing applications.
That's not quite true. Win32 apps that use CreateDialog use absolute positioning for widgets, but the "dialog units" used for that purpose are specifically defined to scale with DPI and font size (one DU is defined as "average width and height of characters in the system font").
We had layout managers in the 90s, perhaps not in the 80s.
> I guess one challenge with this is that old school Win32 doesn't really support anything resembling automatic / responsive layout.
It gives you all the building blocks to implement it yourself.
Remember that Win32 is not UI framework but OS API which allow you to build yourself one (wxWidgets, QT, SWT).
> Now it could be done to maintain backwards compat and have everything completely up to date, but the juice is not worth the squeeze and there are more important things to focus on
You state that as fact, but I believe the poor UI situation is holding windows back from gaining market share. It’s not just the way things look, but uniformly communicate a single experience. Microsoft used to be very integrated and had an entire ecosystem that was uniform. Letting this go has really hurt the brand IMHO, and creates a very low bar for third parties.
The ecosystem was never uniform. In the 95/2000 days there was a lot of 3.11 UI everywhere, in the 7/8 days a lot of XP UI. Office always had its own UI conventions, often being the playground where new ideas were tested.
It's not so much that microsoft let things go as it was taken from them. When computers became networked we needed ways of easily distributing applications across that network in secure, reliable and always up to date ways. Windows never had a good solution to that (not even today), because every install was fraught with peril, and every app had to roll its own update mechanism. Meanwhile the web was sandboxed and up to date by definition, dramatically lowering the barrier to entry for networked software. Anything that could move to the web did move to the web, which opened the door for chromebooks, which in turn fragmented the OS market, which made it uneconomical for companies to invest in a windows native codebase. Microsoft could have stopped this had they leaned hard into the concept of an app store and sandboxing around 2000, and hadn't stumbled so badly when it came to extending windows to mobile.
I'm pretty sure that modernizing the entire windows UI doesn't change a thing for windows market share. But the reason to do it is not because it makes economic sense, it is because they should have pride in the products they make, and want them to be well-made.
> I believe the poor UI situation is holding windows back from gaining market share.
I'm having a hard time believing that. While some Mac users who are used to a more integrated user experience might be deterred, I'm not sure that the os's ui esthetics really affect windows adoption. My feeling is that most people who use Windows do so because they're either (a) told to use it or (b) they choose to use it because it's the shortest path to doing what they want to do.
I'm sure Microsoft know this and understand the tradeoffs very well
I use it because it supports any bit of hardware thrown at it, with little fuss.
I tried to go all Linux many times, gave up, I just wanted something that works.
3 replies →
Most people use it because most people use it. It's what's compatible and widespread, and comfortable and known. It's the reason that people send docx files.
1 reply →
> I'm sure Microsoft know this and understand the tradeoffs very well
Microsoft understands keeping their customers. That is why backwards compatibility is king with Ms.
Windows is not growing in adoption. It is under attack by Google (Android killed Windows on phones and Chromebooks are eating away the education market), Apple, and on the server, Linux.
Windows user from 3.1 to Win 7 here. Once I saw the MacOS experience in 2012, I never looked back. an engineer might say the juice isn't worth the squeeze, but to a consumer, and a power user, the swath of UI and UX inconsistencies is what keeps me away from Windows.
15 replies →
What market share is left there to gain? Ok 17% Mac OS, maybe 5% would switch to Windows?
Strategically, worrying about the low end “worse” competitor is more worthwhile than the higher end competitor. Eg IBM were not undone by a better mainframe.
In Windows’s case the “worse” competitor is ChromeOS and Chromebook, and Microsoft’s basically thrown in the towel when it comes to native apps, so with time everyone who doesn’t need pro Photoshop or AutoCAD could just use a Chromebook.
Note that you can run Word for Android on Chromebooks full-screen already.
8 replies →
That would be for the USA or Canada. Everywhere else, macOS is perhaps a rounding error. ChromeOS gains some traction and is probably more popular than macOS in the education market at least. Windows is still king at least in the world we consider as somewhat developed.
Maybe is not about gaining but stop losing users to MacOS and Chrome os.
And one day Google will release desktop Fuchsia.
I think this inconsistent experience is what is opening the door for competitors like Apple and Linux.
12 replies →
> I believe the poor UI situation is holding windows back from gaining market share.
Big companies just love to jack around with Windows, and do things like prevent users from changing their desktop backgrounds (in the name of "security," of course), and Microsoft has always bent over backwards to give them the ability to do any stupid thing some corporate IT drone can think of, so they continue to be loved my companies around the globe.
There's a _fundamental_ difference in the way that Microsoft and Apple have approached making an operating system. One treats it like a layer of software in an entire stack that someone else owns and controls, and the other makes PERSONAL devices. And it shows.
And, of course, the Microsoft-slobbering trade press can't imagine a scenario where they don't try to prove that YOU should be running Windows because the numbers -- inflated by the Fortune 1000 -- says it's what "everyone" is running. Meanwhile, more than half the people I know now run NOT-Windows for their personal computing.
Let's break out the corporate purchases from sales data, and then let's talk about actual, personal market share, and whether better UI/UX would help Microsoft in a heads-up battle against macOS or ChromeOS.
I have been in the corporate environment for more than 6 year more on the administrator/ engineering side of things. Currently, I am about a year at OrgPad - a very small and quite alternative startup. Two extremes so to say. ;-)
In my experience, Windows is still the target for most specialized or business software. Most of these things also demand a Windows Server component. Typical suspects are: "Personalmanagement" timesheet software basically (accounting working hours, vacation, bonuses etc. and making sure the right receipts are printed at the end of each month and more, like when somebody leaves the company, is ill or the is "Kurzarbeit" because the union with the employer settled on this). Another suspect is software for insurance and taxes, energy management (which in a stainless steel foundry is an important piece of software and keeps at least a handful of people fully employed), logistics software (usually also with needle printers that are able to still print "Durchschlag" carbon paper https://en.wikipedia.org/wiki/Carbon_paper which is fast and handy for the drivers). There is special software for measurement and production with complex machines, ovens, mills, spectrometers, microscopes, x-ray machines etc. Oh and then there is the internally developed application, that keeps running the company and will be replaced sometime during the next 10 years with a different Windows native app if it is ever finished. And there is the MS SQL administration in addition to group policy for all the Windows computers everywhere that also is best done from a Windows computer.
People are used to Windows, older generations in middle and slightly eastern european countries (former east Germany/ DDR included) especially. There just were no Macs in any company besides maybe a top design studio in the capital or regional hub. This is a big factor. In most companies, ordering certain behaviour just doesn't work in practise.
In companies, every purchase is oriented towards "Abschreibung" or the amortisation period that is determined by "Abgabenordnung" if I am not mistaken - basically the tax. To be most efficient on paper, you have to use the device at least for the duration of the period determined by the state tax books, else you will have a bad tax audit etc. Anything that hasn't followed the rule will raise an eyebrow and somebody will have to explain otherwise you will likely pay more in taxes or whatever. I don't understand this stuff at all but I was told it works something like that and therefore you do so much accounting stuff such as inventory of basically everything that costs at least X, every piece of software, where you also get Microsoft and Oracle and other audits from time to time...
If you are not in the business of selling/ writing enterprise or specialized software in some capacity, you don't really have to care. Most things work using a web browser nowadays, even some enterprise software some companies use. Some software is just more performant or can access certain APIs that aren't available in the web browser. Also, it is very hard to support anything for 10+ years in the webbrowser. Stuff just breaks too frequently on the web, even compared to Windows 10.
In the end, people usually buy Windows, because it really is the only platform that actually works for 99,9% of things. You can spend a long time to support the CEO's Mac, because it is the CEO. Of course, most things running a web app or supporting infrastructure can be Linux or UNIX-based without an issue, but the OS on people's desks just is Windows and it will stay like that for a long time. There are few companies that have the resources to support something else on the desktop as well. I bet, these are in the software or hardware development or creative/ design/ art business and not much else.
Education is a very special market, really anything goes there. Simple things tend to win, but what simple is changes school to school. :-) There are schools that it seems don't know what to do with their money and have dedicated IT department, supporting anything Apple, Google, Adobe, AutoDesk and Microsoft is not a problem there. There are schools that cannot even equip one IT classroom and students have to share computers each with a different OS and updates barely work because the internet connection is flaky.
For the first half I thought it was sarcasm.
> creates a very low bar for third parties
The bar for third parties was always nonexistent. That was precisely the advantage of Windows which permitted the boom of development for it in the 90s–2000s.
Holding Windows back from gaining on their 90+% market share?
> It’s not just the way things look, but uniformly communicate a single experience. Microsoft used to be very integrated and had an entire ecosystem that was uniform
Their Office suite may have been internally consistent, but I did a bit of software for Windows 2000 (back when it was the latest version), and I found the UI of Windows (in general) inconsistent, scrollbars would e.g. have subtle difference between applications.
I ended up using Qt for my own work, as I couldn’t figure out what was the official / dominant style to follow.
All this, plus the fact that Win10 has become increasingly unstable for me in the last 6 months (random restarts, nasty performance problems with WSL2 etc.) that it simply isn't worth the money anymore, unless you have some apps that only run on widows (e.g. games :))
Honestly, it sounds like you have faulty HW.
1 reply →
I hope you jest. Windows has far bigger problems than the uniformity of their UI - there's the whole forced update thing, the whole "we really need to spy on literally everything you do on our OS" approach, online advertisement built and integrated with the OS (based on the spying) ... these are the major issues that they need to fixed urgently if they wish to retain their userbase. Windows has been a continuing shitshow since Windows 8 ...
I tried win 10 recently on older hardware with 2gb ram. Long story short: totally unusable. Turned on Win 8.1 on machine with 2gb ram. That was pretty usable and almost pleasant by comparison. I used to think that distro was a stinky turd. The UI in 8.1 was very inconsistent and gave me hives. 10 has changed markedly already between big updates.
The juice is absolutely worth the squeeze. Everybody and their grandmas think Windows is a mess, yet theres no real alternative for most people, so there’s no real incentive for MS to make it better (or for a start, stop making it worse with telemetry, forced restarts and adware).
From MS' point of view I doubt it is worth thr squeeze. You have to consider that the average user and their grandma make very little revenue for them. They buy a license with their computer and that's mostly it. Pro users, companies are a big part of the Windows revenue and for them, backwards-compatibilty is the biggest sales point. I agree this is super bad and confusing for many users, but the majority of them probably never go past the latest design Win10 screens.
> yet theres no real alternative for most people
I just don't buy that. Everyone I've ever persuaded to buy a Mac or iOS device comes to prefer it to Windows in a couple of weeks, and I stop needing to fix something for them every other month.
I actually did that for my parents and in-laws and it went as you said. But there’s no alternative to Windows for a lot of professionals and gamers and that’s an issue. There’s no need for Windows to be better because they own that segment. Also, as much as Windows Desktop dominance sucks, Linux just isn’t there yet and I don’t want to see the walled garden dystopia that a majority of Mac desktops would bring us.
Stating that the UI consistency issues are due to backwards compatibility is stretching the truth to the point of outright deception.
Backwards compatibility is not what stops Microsoft from updating the GUIs for their own components. It is not the reason that the components that have been updated failed to carry forward 100% of the features of the components they're replacing, necessitating the ongoing use of the legacy components. It is not the reason reams of GUIs have gone untouched for 20 years. It is not the reason Windows now has at least 4 versions(!) of Performance Monitor, all of which are broken in at least one glaring way. (More on that below.)
The believable reasons for the GUI inconsistencies I've heard are:
1) API churn, because of which the UI teams did not have sufficient time to work on features versus playing catchup.
2) Bad system GUI APIs that are very difficult for even in-house teams to work with.
3) Unwillingness to take ownership of legacy code, with literally noone left at Microsoft willing to touch things like the ODBC connections panel or the Component Services snapin. People prefer to add new things instead of fixing or removing old things because it's "easier".
Actually, let's just stop here for a second. That last point explains the performance monitor views. I mentioned at least 4 copies, written over decades, by different people. Each new team has steadfastly refused to touch (or remove!) the old code, but hasn't replaced the functionality of the old code, so now end-users need 4 different versions to get things done.
These versions are:
1) The performance tab in Task Manager. The only GUI-based view that exposes some metrics such as GPU temperatures. Only shows a small number of fixed metrics.
2) Resource Monitor, which is opened from Task Manager. The only GUI-based view that shows certain per-process metrics, such as the names of files being touched, or per-connection network stats. Has permanent UI issues that will never be fixed, such as not using the system number formatting in some places, making large metrics unreadable as they change faster than users can count the digits. Similarly, the graphs take forever to change their vertical axes, making them useless 90% of the time.
3) The Performance Monitor MMC snapin. Totally legacy, with un-resizable controls that cut off text. Nonetheless, it is absolutely essential because it provides the only live GUI view of 100% of the performance metrics available in the system. It is also the only way to record metrics and view recordings. It is the only GUI for creating metric logging that persists. Etc...
4) The various versions of Server Manager's performance views, which are so useless that I've literally never used it. Nobody can get their job done doing this, we're all still using RDP to connect to servers so that we can simultaneously launch the three tools above. Why RDP? Because 2 of the 3 above do not support remoting.
I could go on and on like this for hours about how bad just this one aspect of Windows is, let alone the hundreds of other GUIs that have been butchered by bad decision making and internal NIH syndrome leaving a trail of half-baked messes behind.
Wait... did I say 4 performance metrics GUIs? I meant 5, because there's also the new Windows Admin Center, which was clearly written by people that have never had to diagnose a performance issue on a server. It's very pretty and utterly useless, which means that: Nobody will use it, and it will be superseded by someone else's half-baked attempt in a few years, 100% guaranteed.
This may explain why Microsoft have created so many email clients and none of them 100% work.
There are also Microsoft Entourage for Mac OS, few different apps for mobile, and at least two web clients, Hotmail and Outlook.com. Did I miss any?
Microsoft Mail is a product from the early 90s for DOS/Classic Mac OS. Microsoft Internet Mail was a logical upgrade, then added support for newsgroups. Eventually, after the success of Outlook in the MS Office world, it rebranded to Outlook Express. You could still see echoes of Microsoft Internet Mail and News if you looked hard at Outlook Express, its executable was called msimn.exe.
Windows Live Mail and Windows Mail were essentially newer, Vista-era versions of Outlook Express. It's the same program rebranded. Mail was bundled with Vista, Live Mail was a way to push the Live branding and get the newer program into the hands of those running older versions of the OS (WinXP).
Microsoft Outlook is a different beast with a different lineage, part of MS Office, and can be thought of the "pro" version of the basic mail clients bundled with Windows.
Yes, Microsoft could probably do with some branding discipline but the technology is quite predictable.
I'm not sure what Mail.app is, I always thought that was Apple's. Entourage was IIRC a native 'designed for Apple' PIM that eventually made its way into Office for Mac, eventually they decided to rationalize and have one client -- Outlook.
Hotmail / Outlook.com - Outlook.com, again despite the confused branding, is Hotmail evolved. Over time, Microsoft slowly merged its consumer (Hotmail, Windows Live) UI with its 'pro' UI (Outlook for desktops, Exchange Web) so that these days they are all pretty similar. These days you also have Outlook's mobile apps for Android and iOS and those too have a familiar look and feel.
I don't see any of these as a bad thing. Microsoft's history is linked to the history of personal computing and the ebbs and flows of market forces that shaped the PC biz and continues to shape today's technology. Given all the churn in this space, it's actually kind of awesome how their technology has evolved -- and will continue evolving, e.g. with all the focus on a web version of Office. But yes, their sudden shifts in branding is pretty sucky and they could do a lot better there.
But "none of them 100% work" is pretty harsh. No mail client is right for every use case (except emacs, naturally -- if it doesn't work for you it's because you haven't written enough elisp yet). But these are very widely deployed products and do work for a good segment of their target market.
1 reply →
Why Outlook does not work?
2 replies →
>Stating that the UI consistency issues are due to backwards compatibility is stretching the truth to the point of outright deception.
I am not familiar with Windows APIs but I can see this be the truth if MS made it possible for developers to plug inside this dialogs, like a printer driver would plug soem GUI changes in the panels , changing this would break a lot of devices.
Sure, and hence the logical thing to do is to create the new GUI side-by-side with the old one. That doesn't explain the non-extensible GUIs being left behind, or why the new uplifts are so piecemeal, or why they randomly leave out critical features.
1 reply →
Additionally, this is all the more reason why Microsoft should think through before they release a new UI frameworks and instead of releazing and then shortly moving on to a new one. I do not really understand why this cannot be thought through to make it long-lasting without running into alleged backward compatibility issues.
>It is also the only way to record metrics and view recordings.
Does logman not working anymore? If not, hasn't it been replaced by PS APIs?
I did specify that this is related to GUI approaches, not command-line tools. Thankfully the "churn" in the CLI is much lower, but not zero. Most users are happy about the the migration from the legacy CMD-based tooling to PowerShell, but the latest crop of tools are abandoning that and reverting back. E.g.: "dotnet", "bicep", "aks", "az", etc... are all going back to the UNIX-style parameters for consistency with Linux, even though PowerShell is a demonstrably superior shell, especially on Windows.
So even there, the unnecessary and counter-productive interface churn continues to the detriment of the users.
> such as making the OS work/build in such a way that there won’t be inconsistencies in the future
I'm always very skeptical of refusal to handle the past because of the promise that the same won't reiterate in the future. What could also happen is that 5 years from now, you or another dev justifies refusal to handle inconsistencies developed "temporarily" and or accidentally in the meantime with exactly the same reasoning.
A more direct and simple observation is that if nobody is working on converging / cleaning up stuff, they don't converge and aren't cleaned up.
As for the amount of backward compat already provided by Windows, it is good but not great anyway. Try to play old games, install/use old VS with a few patches, install/old Matlab, use old devices, etc... Hell even try to continue to use your semi-recent devices across new build of windows, like 20H2 breaking USB-C on my XPS 15 9560 (but now I'm ranting so I'll stop :D )
Is this also the reason why there are no tabs in windows explorer?
I'm a TreeStyleTab user in Firefox with always >200 Tabs open. But I never got the idea of Tabs in Explorer (or Finder for that matter).
When I use multiple Explorer Views, most of the time I do it, to interact between them. This is not the case for websites.
So for drag and drop interactions, two windows always beat the drag, hover over the other tab until it catches my intend to switch to this tab, and drop it there workflow.
The same goes for copy and paste. Ctrl+c, alt+tab, ctrl+v. I'm sure there is a corresponding command to cycle through the tabs, but alt+tab is deep muscle memory, that it beats this command always.
> I'm sure there is a corresponding command to cycle through the tabs, but alt+tab is deep muscle memory, that it beats this command always.
It's ctrl+tab. And shift will reverse direction in both. It's nice and related.
And then there's MacOS, where cmd-tab doesn't cycle through windows, but through applications. Cmd-` cycles through windows of the same application. Because apparently, using multiple applications and multiple windows means you're a bad boy and need to be punished.
1 reply →
Generalized tabs accros even different programs is an experiment that has been attempted a few years ago in a few insider builds, but has been quickly abandoned. IMO it was too complex from a UX pov, duplicated some of the use cases of virtual desktops, etc. Maybe had they tried tabs for single applications, it would have succeeded. In this alternate universe we may not have the new Windows Terminal program though (IIRC adding tabs specifically in console.exe was prevented because they were about to start this grand inter-program tab design instead, and Windows Terminal was started after the experiment failed)
It is a power user feature which most users won't will not appreciate. I am using Mac for about 7 years now, and almost never use tabs in Finder.
Tabs are great for saving screen real estate. Finder windows are relatively small and, with large screens, the need to cram two or more folders into the same window ceases to exist. It’s easier to remember the window to the right is my docs folder and the one to the left is the downloads one.
2 replies →
Honestly thank goodness for the backwards compatibility, otherwise I'd have to rewrite an existing legacy app I've been put in charge of maintaining at my place of work. The thing is in winforms and a weird third party UI library built with winforms as well. It would just be a nightmare for me to have to go through each screen to recreate it reliably.
They tried modernizing with Windows RT, which was actually very smooth and functional if it wasn't so locked down.
Wasn't RT just windows 8 with an ARM build target and win32 stuff removed so they could sell cheaper tablets?
So if you wanted an "unlocked" windows RT, couldn't you just have run windows 8?
>win32 stuff removed
Win32 was not removed from Windows RT, but your access to it (if your name wasn't "Microsoft"; because that low-level access was reportedly how they got Office on it) was.
And because of Secure Boot, which (unlike x86, for now) is mandated enabled for Windows on ARM systems, you had to wait for an unpatchable security hole in the booting process (which, while it exists, wasn't relevant until past the end of the Surface RT's lifecycle), you couldn't run full Windows 8 or a tablet-first Linux distro.
1 reply →
This is exactly what unit tests are for. Besides, the business logic should be already fairly separated from the presentation logic.
As for preventing inconsistencies in future, try getting your QA team back.
Could you ever see Microsoft dropping the registry?
My probably unpopular opinion is that the registry is the worst feature in Windows. It’s bloat directly impacts the speed of the operating system.
Seems like such a waste for so many advancements in Windows but the registry sticks around like a bad hangover.
It is more on the contrary. The registry is one of the most well-thought features of Windows.
Also there is nothing like "bloat". No process needs too loop around or smth like that, so there is literally no way to have a forgotten registry item slowing down your PC.
And last but not least: What are the alternatives? .plist config files like with Apple, scattering around the system?
If I remember correctly there was an internal effort at Microsoft to scrap the registry (Longhorn?) but they decided not to, and probably because a registry is actually a good idea and necessary to get something performant (compared to ini files).
The critique against the Windows registry stems from the Windows 95 days when the registry could become corrupt and repair or reinstall was necessary. However since then the registry works like a database, it has transaction logs and simultaneous writes are atomic. Registry is also strongly typed, not a bunch of strings. Registry can easily be backed up and restored.
Probably because Longhorn tried to replace even more parts of Windows with a database-like system, and they'd have rolled the registry-equivalent into that, if it had succeeded.
> It’s bloat directly impacts the speed of the operating system.
Do you have any evidence of this? I remember in the XP days the German c't computer magazine tested populating the registry with hundreds MBs of garbage and found no performance impact at all.
How is it worse than having config files scattered around in distro-specific locations?
And how is it worse than dconf? People hated on dconf during the transition to systemd, but it is here now.
I do not really think one is better than the other.
1 reply →
As other comments point out, there is nothing wrong with the registry itself. On the other hand the editor (regedt) is terrible.
If you drop something, then you need some kind of replacement for that.
I don't see any alternative on where entire OS could be configured other than registry.
KDE and all its applications and a majority of other Linux desktop application use the configuration laid out in <https://specifications.freedesktop.org/basedir-spec/>.