VOGONS

Common searches


When will Microsoft finally die?

Topic actions

  • This topic is locked. You cannot reply or edit posts.

Reply 100 of 122, by vladstamate

User metadata
Rank Oldbie
Rank
Oldbie

I really don't see a purpose for this thread. It superimposes a person's personal preferences and ideas over a multi-hundred-billion dollar company and industry while at the same time asking fundamental questions about when it will die. A simple look at the financials (including sales, shares price, market spread) will tell you "Not anytime soon." This is just devolving into "Win7 is way better than Win10" or "OSX is just as good as Linux" for which we can never hope to all agree to. This is like opening threads saying "I don't like company X. When will it die?"

YouTube channel: https://www.youtube.com/channel/UC7HbC_nq8t1S9l7qGYL0mTA
Collection: http://www.digiloguemuseum.com/index.html
Emulator: https://sites.google.com/site/capex86/
Raytracer: https://sites.google.com/site/opaqueraytracer/

Reply 101 of 122, by keenmaster486

User metadata
Rank l33t
Rank
l33t
vladstamate wrote:

I really don't see a purpose for this thread.

Honestly, I must admit this thread was kind of troll-y. I apologize.

I think there can be an honest and civil debate though - not about "when will MS die" but about the following, which I have gleaned from the contents of the thread so far, and which I invite others to criticize:

1. There is no "perfect", "universal" PC OS out there right now. Windows has the most support but also has bugs and is privacy-invading, not to mention it being the most popular OS which invites a great deal of malware. OS X also has a lot of support but involves large quantities of cash outflow and having to do everything exactly the way the ghost of Steve Jobs says you must. Linux-based OS's are the most inherently stable and well-designed, but suffer from lack of support and, still, a little "please be a power user" syndrome.

2. Given this, what do we think the future might hold? Will there ever be a "perfect" OS (for all practical purposes)? Is MS converging on the right solution? Is Apple? Will there emerge the perfect Linux distro?

I don't know the answer to that question. I have ideas of my own though.

If I was a billionaire I would start my own computer company making well-designed Intel-based machines and focusing on developing our own Linux distro and accompanying software which would be distributed for free. The money would be made on the hardware, which would obviously have to stand out to compete with existing manufacturers. The "big sell" marketing-wise would be about the same as Apple (software works perfectly with hardware, everything "just works", great, intuitive interfaces, and sleek but powerful hardware), except with the added aspect that everything software-wise is "free". Having the resources to lobby companies like Autodesk to support Linux would be a must.

World's foremost 486 enjoyer.

Reply 102 of 122, by gdjacobs

User metadata
Rank l33t++
Rank
l33t++
Scali wrote:

The important thing there is not the absolute figures or exact scenarios, but learning the mindset of working inside certain limitations and pushing the boundaries.
People who ask "Why does it matter" will never make good engineers.
Real engineers know that the answer is: "Because we can!"

There's lots of answers:
"Because it will be handy!"
"Because it will be cool!"
"Because it's the most efficient!"
"Because it's the most straightforward!"

Probably lots of others.

keenmaster486 wrote:

Linux-based OS's are the most inherently stable and well-designed, but suffer from lack of support and, still, a little "please be a power user" syndrome.

I'm a big Linux fan, but I would contest this assertion. Between different applications and different distributions, there's tremendous variability in any number of meaningful performance aspects. It's hard to make such a blanket statement with so much diversity without losing important nuance.

All hail the Great Capacitor Brand Finder

Reply 103 of 122, by SquallStrife

User metadata
Rank l33t
Rank
l33t

Linux-based OS's are the most inherently stable and well-designed, but suffer from lack of support and, still, a little "please be a power user" syndrome.

I'm a big Linux fan, but I would contest this assertion. Between different applications and different distributions, there's tremendous variability in any number of meaningful performance aspects. It's hard to make such a blanket statement with so much diversity without losing important nuance.

It's as ridiculous of a statement as everything else he said. (Or maybe doesn't know what "inherently" means?)

There are many qualities and attributes of GNU/Linux that are cobbled together and shoehorned, and aren't "well designed" at all. Audio is a great example, the abomination that is the X windowing system is another. They're designs and architectures that have carried over from the 70s and 80s, but nobody's game to change. Wayland has been around for 9 years, and hasn't made any serious headway. How many competing audio stacks are there still? All similar, none perfect...

Meanwhile Windows and macOS have had excellent low-latency unified audio stacks for years.

For the record, computers running Linux outnumber those running Windows in my house when you consider Raspberry Pi's and other embedded systems. I love Linux. I just can't romanticise it. It has flaws. Huge flaws. All operating systems do.

VogonsDrivers.com | Link | News Thread

Reply 104 of 122, by gdjacobs

User metadata
Rank l33t++
Rank
l33t++
SquallStrife wrote:

There are many qualities and attributes of GNU/Linux that are cobbled together and shoehorned, and aren't "well designed" at all. Audio is a great example, the abomination that is the X windowing system is another. They're designs and architectures that have carried over from the 70s and 80s, but nobody's game to change. Wayland has been around for 9 years, and hasn't made any serious headway. How many competing audio stacks are there still? All similar, none perfect...

Part of the reason why I really dislike Poettering. Pulseaudio has been broken to various degrees since inception. Rather than fix it, Lennart has decided to spend his time writing bad replacements for mission critical components.

I moved to Jack as my primary sound interface and haven't looked back. It's dev base is presumably satisfied without adding the few convenience features PA still holds over it, but I suspect the boffins at RH would proceed with the necessary modifications if they didn't suffer so heavily from Not Invented Here. Of course, Ubuntu is too busy having their own adventures with cell phones, app stores, and everything else.

Wayland will get there eventually. At this point, it feels like it's being held back by downstream projects completing their transition. However, I doubt the difference in functionality will be that huge. X does provide compositing - Wayland will provide a welcome cleanup while providing a more battle tested compositing system as DEs will no longer have to roll their own. Meanwhile, Ubuntu will still be spinning their wheels with Mir.

SquallStrife wrote:

Meanwhile Windows and macOS have had excellent low-latency unified audio stacks for years.

I understand MacOS has a nice audio stack. I don't have a personal testament to that, as I've never really used it heavily for that purpose. In Windows, I've been less than impressed. I've found ASIO drivers to be extremely variable in quality, although it might be a function of the audio devices I've used.

All hail the Great Capacitor Brand Finder

Reply 105 of 122, by dr_st

User metadata
Rank l33t
Rank
l33t
gdjacobs wrote:

This is a false dichotomy. The question is not efficient vs inefficient. It's a question of how you weigh the cost (and benefit) of an added feature. For instance, I don't give a crap about Windows Live integration. I'd probably want it gone and maybe save a bit of RAM resident size in the form of hooks to the DM, but for Microsoft it's a good thing so it goes in.

These are really different aspects of the same thing. Choosing which features to include is also a hard decision in the presence of limited resources. When the resources are abundant, you can "go nuts" and include everything and the kitchen sink. You don't care about Windows Live, but apparently some folks do, and there are probably features that you care about and others don't, and they are also included, because the resources allow it.

AlaricD wrote:

We do, but what programmers are really using the SSE4 instructions (that is, if they're using assembly anyway)? What compilers can take high-level code and truly optimize it for a Core i7 wherein it can't run on a Core 2 Quad by virtue of using any of those instructions?

The Intel Compiler is said to specifically be able to optimize high-level code for specific Intel CPUs.

AlaricD wrote:

I just don't see that the Windows operating system should *require* a Core iX processor, since there's really no compelling reason for that.

What Windows operating system requires a Core iX processor?

AlaricD wrote:

It seems that sometimes programmers use "the computer is THIS powerful anyway" to be a bit sloppy in their code.

It's fine. any time not spent optimizing tiny bits and pieces for maximum efficiency can be better spent solving actual problems users want solved.

AlaricD wrote:

I recall a programming class years ago, wherein the instructor was giving all kinds of ways to even reduce the complexity and memory/storage footprint of the source code, and someone asked with some derision "why does this matter? Everyone's got Pentium 100MHz processors and 8MB of RAM so why go to that extreme?" Guess the guy never had to constrain himself to ~37KB with 6502 assembly on a C64.

He's right (and more right today than he was back then). It doesn't matter.

Unknown_K wrote:

I don't like Windows 10 because a perfectly good Win7 video card is not supported in 10. Re inventing the UI every so many years is NOT a good idea when the old ones works fine. Oh and having to deal with just stereo working on a newer sound chipset that works just fine in Win7.

In 99% of the cases that's a problem of the hardware vendor. Microsoft went to great lengths to make sure that Win10 is as backwards compatible to Win7 in terms of drivers and API as possible, and asked vendors to ensure that Win7 drivers just work on Win10. Not all manufacturers listened.

Often a driver not working on Win10 that worked on Win7 is just due to a mechanical restriction in the INF file, which the hardware vendor added either on purpose or due to stupidity. If you remove that restriction, everything 'magically' works. I've had this experience with one or two Intel integrated GPUs.

https://cloakedthargoid.wordpress.com/ - Random content on hardware, software, games and toys

Reply 106 of 122, by Scali

User metadata
Rank l33t
Rank
l33t
keenmaster486 wrote:

Linux-based OS's are the most inherently stable and well-designed, but suffer from lack of support and, still, a little "please be a power user" syndrome.

Excuse me, but linux is one of the worst-designed modern OSes out there.
Have you never heard about the Tanenbaum vs Torvalds discussions for example?
Linux pretty much breaks all the rules of proper modern OS design, or of sound software engineering and architecture in general (basics such as modularization, layered design, stable interfaces etc).

Windows NT on the other hand has a solid technological basis, and the only corners they 'cut' design-wise are done for practical reasons: doing everything 'by the book' may result in performance that is orders of magnitude slower in some cases.
But if you look at the evolution of Windows NT, you will see that Microsoft does 'fix' some of these things whenever this is possible. For example, in earlier versions, a lot of performance-critical things were done directly in kernel-space, to avoid too many costly context-switches. In later iterations, where newer CPUs allow more efficient context switching and such, the driver models have been modified to move certain tasks back to user-space where they belong.

http://scalibq.wordpress.com/just-keeping-it- … ro-programming/

Reply 107 of 122, by Scali

User metadata
Rank l33t
Rank
l33t
gdjacobs wrote:

I've found ASIO drivers to be extremely variable in quality, although it might be a function of the audio devices I've used.

It is, and you cannot blame Microsoft in any way.
ASIO is not even a Microsoft standard. It was developed by Steinberg for the first DAW versions of CuBase on Windows, where high-performance, low-latency audio was required.
It merely shows the strength and openness of the Windows design that a third-party can create something like this so effectively.

ASIO drivers are the most basic audio drivers you can think of. They perform no processing whatsoever. They just transfer tiny DMA buffers between the hardware and the application layer (both input and output buffers are done at the same time, to reduce overhead).
Since Windows has an extremely granular pre-emptive design, it can easily process driver events with less than 1 ms latency, without any additional overhead (this cannot be done in linux, the only way to get low latency there is to basically 'hack' the whole OS and use extremely small timeslices everywhere, so that they do not have to be pre-empted. Downside is obviously far more context switches and therefore lots of performance overhead).
As a result, ASIO is mostly dependent on how fast the underlying audio hardware can transfer buffers.
There's some very well-designed hardware out there, which can work reliably with extremely small buffers (of 64 samples or even less), without ever missing a beat, giving you 'realtime' audio processing capabilities (I had a Terratec EWX24/96 in a Pentium II 400 back in the day, that worked great... I might still have used it today if I had PCI slots).
Cheap/bad hardware (please, do not ever use ASIO4All, and say that is ASIO) requires 5 ms or more latency.

Sometimes the chipset can also be a problem. Back in the day I had an Athlon 1400 with VIA KT133A chipset, and I couldn't get the latency below 30 ms on my USB devices.
Then I plugged them into a system with Intel chipset, and suddenly the same USB device could handle 3-5 ms without a problem. So it wasn't the device, it was the lousy VIA USB controller (or perhaps the PCI bridge it was connected to).

http://scalibq.wordpress.com/just-keeping-it- … ro-programming/

Reply 109 of 122, by gdjacobs

User metadata
Rank l33t++
Rank
l33t++
Scali wrote:

It is, and you cannot blame Microsoft in any way.
ASIO is not even a Microsoft standard. It was developed by Steinberg for the first DAW versions of CuBase on Windows, where high-performance, low-latency audio was required.
It merely shows the strength and openness of the Windows design that a third-party can create something like this so effectively.

I seem to recall that. However, it has become the de-facto standard for low latency pro audio in Windows, so by extolling the virtues of the Windows sound API, one is effectively praising (at least partially) ASIO.

Just as a point of reference, real world data from three years ago seems to indicate that Linux audio latency is now competitive. Generic HDA sound chips can deliver 48khz audio at 2.25ms latency. USB DACs have a little bit more latency, around 8ms.

All hail the Great Capacitor Brand Finder

Reply 110 of 122, by Scali

User metadata
Rank l33t
Rank
l33t
gdjacobs wrote:

I seem to recall that. However, it has become the de-facto standard for low latency pro audio in Windows, so by extolling the virtues of the Windows sound API, one is effectively praising (at least partially) ASIO.

ASIO pre-dates all current Windows sound APIs though.
It also serves a different market (most consumer-grade sound cards don't even have support for it in the first place, and most consumers will not miss support, because the applications they use don't rely on ASIO).

gdjacobs wrote:

Just as a point of reference, real world data from three years ago seems to indicate that Linux audio latency is now competitive. Generic HDA sound chips can deliver 48khz audio at 2.25ms latency. USB DACs have a little bit more latency, around 8ms.

I wouldn't exactly call that 'competitive'.
The other question you have to ask is: at what cost? Can you get these figures from any default linux distribution out-of-the-box? Because you can with ASIO.
I haven't checked recently, but at least a few years ago, you still had to use special DAW-oriented distributions of linux, which can with a pre-patched linux kernel and drivers. Also, these distributions had higher CPU overhead than regular distributions.

http://scalibq.wordpress.com/just-keeping-it- … ro-programming/

Reply 111 of 122, by Woolie Wool

User metadata
Rank Member
Rank
Member
SquallStrife wrote:
It's as ridiculous of a statement as everything else he said. (Or maybe doesn't know what "inherently" means?) […]
Show full quote

Linux-based OS's are the most inherently stable and well-designed, but suffer from lack of support and, still, a little "please be a power user" syndrome.

I'm a big Linux fan, but I would contest this assertion. Between different applications and different distributions, there's tremendous variability in any number of meaningful performance aspects. It's hard to make such a blanket statement with so much diversity without losing important nuance.

It's as ridiculous of a statement as everything else he said. (Or maybe doesn't know what "inherently" means?)

There are many qualities and attributes of GNU/Linux that are cobbled together and shoehorned, and aren't "well designed" at all. Audio is a great example, the abomination that is the X windowing system is another. They're designs and architectures that have carried over from the 70s and 80s, but nobody's game to change. Wayland has been around for 9 years, and hasn't made any serious headway. How many competing audio stacks are there still? All similar, none perfect...

Meanwhile Windows and macOS have had excellent low-latency unified audio stacks for years.

For the record, computers running Linux outnumber those running Windows in my house when you consider Raspberry Pi's and other embedded systems. I love Linux. I just can't romanticise it. It has flaws. Huge flaws. All operating systems do.

Indeed, and I think that Linux will never succeed as a mainstream desktop operating system (it has succeeded immensely, and rightly so, in embedded, server, and infrastructural applications) partly because of legacy software that nobody wants to get rid of like X as you mentioned, partly because most users are already committed to Windows software (good luck transitioning my employer's booking which is based on a 12-year-old version of QuickBooks over to Linux...hell, it doesn't even like Windows 10 that much), and partly because of the philosophy and priorities of the Linux community themselves. Even on the most newbie-friendly Linux distro (I don't count bastardized versions full of proprietary components like Android or ChromeOS as Linux), it's easy, by design, for that user to mess with its carefully curated and preconfigured software and introduce problems. That's great for enthusiasts who want to customize their Linux, but Microsoft has other reasons besides greed, arrogance, or malice to make core parts of Windows NT "black boxes"--so Joe End User's dumb ass doesn't start messing with them and wreck his OS and blame Microsoft for it.

dr_st wrote:

Indeed all this is correct, but they don't change the fundamental truth, that you have to have certain level of trust towards the corporation whose device you use. You have to trust Microsoft that when they say that they don't store any personal information collected via telemetry channels, that they really don't. You have to trust Google that when you turn off location tracking in your Android device, it really stops tracking (even though it obviously still has the ability). If you don't trust them, you should not use their products.

I don't agree with this. It is almost certain that both Google and Microsoft are lying about that, especially since governments are legally empowered to force them to turn over personally identifying information. However, part of living in a class economy is that the owners of capital are going to screw you and there's nothing you can do about it except live like a hermit or move to Amish country (many Amish households and establishments have computers, with no multimedia or internet access, to help them manage their farms, businesses, and homesteads, and they never replace them unless they're utterly broken, making them among the first retro computer users 😉).

wp0kyr-2.png CALIFORNIA_RAYZEN
1wpfky-2.png REDBOX
3q6x0e-2.png FUNKENSTEIN_3D

Reply 112 of 122, by gdjacobs

User metadata
Rank l33t++
Rank
l33t++
dr_st wrote:

These are really different aspects of the same thing. Choosing which features to include is also a hard decision in the presence of limited resources. When the resources are abundant, you can "go nuts" and include everything and the kitchen sink. You don't care about Windows Live, but apparently some folks do, and there are probably features that you care about and others don't, and they are also included, because the resources allow it.

How very undemocratic.

dr_st wrote:

The Intel Compiler is said to specifically be able to optimize high-level code for specific Intel CPUs.

Auto vectorization does exist, but it's not universally effective. That's why CPUs have sophisticated mechanisms to schedule, fuse, and dispatch uOps (like RISC for x86 CPU backends) in such a way that the pipeline remains as full as possible.

dr_st wrote:

It's fine. any time not spent optimizing tiny bits and pieces for maximum efficiency can be better spent solving actual problems users want solved.

Sometimes less is more.

dr_st wrote:

He's right (and more right today than he was back then). It doesn't matter.

He's also wrong. What about energy usage? What about latency? What about CPU cost (when ordering by the millions for embedded work)?

dr_st wrote:

In 99% of the cases that's a problem of the hardware vendor. Microsoft went to great lengths to make sure that Win10 is as backwards compatible to Win7 in terms of drivers and API as possible, and asked vendors to ensure that Win7 drivers just work on Win10. Not all manufacturers listened.

Often a driver not working on Win10 that worked on Win7 is just due to a mechanical restriction in the INF file, which the hardware vendor added either on purpose or due to stupidity. If you remove that restriction, everything 'magically' works. I've had this experience with one or two Intel integrated GPUs.

I've had the same experience with Intel GPUs.

His point about two UIs is spot on. The continued project to move to the "modern UI" or whatever they call it this week continues to be a shit sandwich. Windows 10 continues to have two completely different control panels. The new one is so badly designed that it drops important functionality. How is this an improvement?

Seriously, it's like MS decided that they needed a bro coder as a platform architect or something.

All hail the Great Capacitor Brand Finder

Reply 113 of 122, by vladstamate

User metadata
Rank Oldbie
Rank
Oldbie

One question that bears attention is: do we really have the "Joe Dumb" user anymore? Sure back in the 80s, 90s, 00s and so on there were people who could see a computer for the first time and for whom even a Windows Vista interface could have been a bit overwhelming.

But nowadays? In the developed world children as old as 6 year deal with iPhones, iPads, etc (my kids do that) so the question of finding their way around a computer interface is not a new one anymore.

So arguments like: lets make a user interface for an OS that "just works and is easy to use" is not defining anymore. Since usability is a not a hard wall anymore. In 20 years from now even grandmas will know how to interact with a PC. Eventually Linux desktop in general cannot hide behind: well, we are not used as much because our interface is not as elegant and easy to use as OSX or Windows.

YouTube channel: https://www.youtube.com/channel/UC7HbC_nq8t1S9l7qGYL0mTA
Collection: http://www.digiloguemuseum.com/index.html
Emulator: https://sites.google.com/site/capex86/
Raytracer: https://sites.google.com/site/opaqueraytracer/

Reply 114 of 122, by Woolie Wool

User metadata
Rank Member
Rank
Member

Yes, because iThings are not general-purpose computers like a PC--in philosophy they're more like a vastly more powerful and sophisticated version of a Wang word processing machine. It takes specialized knowledge and equipment (and sometimes violating your mobile contract) to even gain root access to a smartphone or tablet. You don't have to know how Android or iOS works under the hood, in fact, unless you have a good reason to (e.g. developing software), Apple and Google would really rather you didn't. GNU/Linux exposes all its internal functions to the user; this is a core and inalienable aspect of GNU/Linux (using the "GNU" part here to distinguish it between other Linux-kernel operating systems like Android) and one of its primary design goals. The problem is not "a computer interface"; it's all the stuff behind the interface. Windows restricts your access to it and Android abstracts it away entirely for 99.9% of users. I have never used a terminal prompt on my smartphone, nor have I ever had a reason to. In GNU/Linux it's all readily available and the design of the OS and the values and practices of the Linux community encourage users to get their hands dirty in it. You can try to use a Linux machine as an appliance, but it is not how Linux is meant to be used and eventually most users will end up at the command line using sudo, messing with things with the attendant risk of breaking something, or go back to Windows.

It's way more fundamental than the what of Linux--software packages, desktop environments, init vs. systemd, it comes from the why of Linux--the people who maintain it, the people who use, the community around it, the values and goals they have. GNU/Linux is an OS by and for hackers and people who love computers, and this philosophy pervades every line of code of every free software package in the entire GNU/Linux ecosystem. Linux doesn't just require you to mess with it--it conditions the user to want to mess with it, because it was designed by a community of people who consider administering, configuring, and tweaking a personal computer a fundamental right and responsibility (remember, rights and responsibilities are inextricably linked--the right cannot exist without the corresponding responsibility) of being a user. No slick desktop environment can get rid of this; it is GNU/Linux's very soul.

wp0kyr-2.png CALIFORNIA_RAYZEN
1wpfky-2.png REDBOX
3q6x0e-2.png FUNKENSTEIN_3D

Reply 115 of 122, by vvbee

User metadata
Rank Oldbie
Rank
Oldbie
vladstamate wrote:

One question that bears attention is: do we really have the "Joe Dumb" user anymore? Sure back in the 80s, 90s, 00s and so on there were people who could see a computer for the first time and for whom even a Windows Vista interface could have been a bit overwhelming.

But nowadays? In the developed world children as old as 6 year deal with iPhones, iPads, etc (my kids do that) so the question of finding their way around a computer interface is not a new one anymore.

So arguments like: lets make a user interface for an OS that "just works and is easy to use" is not defining anymore. Since usability is a not a hard wall anymore. In 20 years from now even grandmas will know how to interact with a PC. Eventually Linux desktop in general cannot hide behind: well, we are not used as much because our interface is not as elegant and easy to use as OSX or Windows.

I think you're confusing the interface coming to the user and the user coming to the interface. Mobile design is toward simplicity and I'd say we have more joes and janes now.

Reply 116 of 122, by dr_st

User metadata
Rank l33t
Rank
l33t
gdjacobs wrote:

How very undemocratic.

How so?

gdjacobs wrote:

How very undemocratic.

How so?

gdjacobs wrote:

What about energy usage? What about latency? What about CPU cost (when ordering by the millions for embedded work)?

If you're building general-purpose software for PCs, then none of this matters. If you building something very specific, embedded or real-time, then of course it still does, and then these questions become important, and if you are good at optimizing on that level, you're in for a great career and a lot of money in these fields.

gdjacobs wrote:

His point about two UIs is spot on. The continued project to move to the "modern UI" or whatever they call it this week continues to be a shit sandwich. Windows 10 continues to have two completely different control panels. The new one is so badly designed that it drops important functionality. How is this an improvement?

It's not necessarily an improvement. I see the UI of Win8/Win10 as something parallel to the core OS, and whereas the core OS really has been getting better with each generation, the UI is not conclusive. However, I will not agree that it's as atrocious as some say it is. It has it's ups and downs. Microsoft is aiming for a unified, universal platform, and as you can imagine - that's kinda hard, so things don't always turn out optimal. Certain things are easier with the new UI, certain harder. In that sense it's been more or less the case with every major change.

From an aesthetic point of view, I do prefer the consistent desktop UI of Vista/Win7. But I don't hate the new one enough to stay off the new versions.

https://cloakedthargoid.wordpress.com/ - Random content on hardware, software, games and toys

Reply 117 of 122, by Woolie Wool

User metadata
Rank Member
Rank
Member
vvbee wrote:
vladstamate wrote:

One question that bears attention is: do we really have the "Joe Dumb" user anymore? Sure back in the 80s, 90s, 00s and so on there were people who could see a computer for the first time and for whom even a Windows Vista interface could have been a bit overwhelming.

But nowadays? In the developed world children as old as 6 year deal with iPhones, iPads, etc (my kids do that) so the question of finding their way around a computer interface is not a new one anymore.

So arguments like: lets make a user interface for an OS that "just works and is easy to use" is not defining anymore. Since usability is a not a hard wall anymore. In 20 years from now even grandmas will know how to interact with a PC. Eventually Linux desktop in general cannot hide behind: well, we are not used as much because our interface is not as elegant and easy to use as OSX or Windows.

I think you're confusing the interface coming to the user and the user coming to the interface. Mobile design is toward simplicity and I'd say we have more joes and janes now.

Indeed. I once read an article by a FreeBSD fanboy that was basically a bunch of complaining about Windows from the 2000/XP era and while much of it was either exaggerated or missing the point, the author did make one astute observation--most PC users of the era did not need a general-purpose computer. They needed...something like a smartphone, a device with deliberately restricted capabilities and a highly abstracted user experience that keeps the nuts and bolts of the system out of the user's sight, mind, and hands. Billions of people use smartphones with no idea how a computer works or how to set up or maintain one. They don't know, they don't care, and mobile devices are designed for them and their needs, and they fulfill those needs very well.

wp0kyr-2.png CALIFORNIA_RAYZEN
1wpfky-2.png REDBOX
3q6x0e-2.png FUNKENSTEIN_3D

Reply 118 of 122, by Scali

User metadata
Rank l33t
Rank
l33t
dr_st wrote:

If you're building general-purpose software for PCs, then none of this matters.

What is 'general-purpose'?
I can think of a number of very common applications, where performance is very important.
One obvious example is games: The main selling point of many AAA titles is how much eyecandy you get, and how well it runs on the average person's system.
Games like Q3A, Half-Life2 and DOOM 2016 are perfect examples of games that are highly optimized to get 60 fps on a wide range of systems, so people can get a very good online experience.
Developing game engines and shaders is all about the optimization. Game developers compete directly with eachother on eyecandy and framerate.

In general, if you develop middleware (libraries and development kits), performance is a very important aspect.

Another example is movie editing software for example. Especially for high resolutions, such as 4k, you want highly optimized post-processing filters and encoding/decoding, because with unoptimized software, it will take forever to create movies even on the most high-end systems you can buy.

There's probably plenty of other examples (VST? The faster they are, the more effects and instruments you can use, without having to mix them down in between).
OSes themselves are obviously also highly optimized. As are many drivers, especially for video cards. Performance is everything.

On the flipside this also means that the average programmer who uses a modern OS, and a modern programming framework, will make use of a lot of highly optimized libraries and modules.
In the old days people had to develop many datastructures and algorithms themselves, such as hashtables, binary search trees etc. These days they are included, either in the libraries, or even as a fundamental part of the language itself.
I would say there just is a lot more well-optimized code around these days than there was in the past.

Of course that doesn't prevent some developers from being completely incompetent, and making terrible software.

http://scalibq.wordpress.com/just-keeping-it- … ro-programming/

Reply 119 of 122, by Cyberdyne

User metadata
Rank Oldbie
Rank
Oldbie

Microsoft is the industrial leader, it has given us some great software even i have hated Microsoft software, but in reality i love my Windows 7 main machine, and i absolutely love Windows XP, and my retro machines that have Windows 98SE or Windows NT 3.51 or Windows 3.11 FWG. Those are the best.
And in reality Office is not that bad either, well if you use 4.3(16 bit) or 97(32bit). And i have tried all DOSses and still like MS 6.22 and 7.10 the best, ok DR has it moments and IBM 2000 has lots of tools, but that about it.

PS. In reality we do not need to talk about Windows Millenium, Vista, 8.0 8.1 and 10.

I am aroused about any X86 motherboard that has full functional ISA slot. I think i have problem. Not really into that original (Turbo) XT,286,386 and CGA/EGA stuff. So just a DOS nut.