MB Portability, Customizations, The Core

[alexei]

There are two areas where ReactOS should instantly overcome in competition with MS:
Portability (in terms of ability of the same installation to survive moving to a different motherboard or VM) and
giving users ability to make custom lightweight OS for specific needs, such as Car-PC, OS on usb, etc.
MS created its famous 7B BSOD and didn't let XP Embedded on desktops for obvious business reasons.
ReactOS can become the solution to all who's having hard time removing MS crap with nLite, etc.

I consider to be highly important to make the OS core not depent on all that "fancy stuff" that came aftre NT4 (I'm not talking about USB support and other new hardware, though). Unfortunately, along with improving user interface MS was adding non-removable "features" to each subsequent version of windows making them worse in each release.

People seem forgetting what they new decades ago: OS should have several levels of functionality, and lower levels must not use functionality from higher levels. It may be an interesting to discuss what should be considered the core functionality of ReactOS?

[Lone_Rifle]

What you describe as fancy stuff are entities that things that came out later (post-NT4 drivers, games that use later versions of the DirectX runtime, applications that depend on .NET frameworks) probably will be dependent on. If you want to confine your user base to using old pre-NT4 drivers and such, please feel free to fork the project. ReactOS is meant to be a drop-in replacement for Windows, which can readily run applications and drivers written for the latter, and people on this forum seem to always overlook this when advocating implementing/removing things that would break compatibility.

In addition your notion of Microsoft operating systems having 'non-removable features' in the user interface seems to indicate that you have no idea how to disable the Themes service in the NT5 generation of operating systems, and/or your ignorance of the presence of the Windows Classic interface alongside Luna and Aero (any arguments that the latter two take up hard drive space can be readily dismissed by the fact that space is cheap anyway). The rest of what you said is already common knowledge amongst the developers, and merits no further discussion.

[Haos]

who's having hard time removing MS crap with nLite, etc.

Just because you dont use it, or rather, have no knowlegde how to use it, it doesnt necessarily has to be crap... Mind that please.

[alexei]

By "removal" I meant complete exclusion of all related components, not disabling their functionality.
It may be hard to believe, but many people still don't want useless (to them) files on their hard drives, regardless of how much fee HDD space they have. Defragmentation time, size of backups, VM size, etc. are still important, and SSD are still expensive.

Regarding drivers, yes, that's how MS built its monopoly. Sad story.

Eventually, ROS forks will come, my major point was that developers may try minimizing cross dependensy of ROS components and provide better base for its customization. Even MS finally came to understending that minimal core has significant value. Read this http://www.vnunet.com/vnunet/news/22016 ... ni-windows

ReactOS is meant to be a drop-in replacement for Windows, which can readily run applications and drivers written for the latter, and people on this forum seem to always overlook this

Probably there is a reason for that. Probably ROS interests those who's dissatisfied with MS policies and practices. "Windows enthusiasts" are looking at ROS in hope to get compatible alternative, not a free drop-in replacement to save some money.

[Lone_Rifle]

By "removal" I meant complete exclusion of all related components, not disabling their functionality.
It may be hard to believe, but many people still don't want useless (to them) files on their hard drives, regardless of how much fee HDD space they have. Defragmentation time, size of backups, VM size, etc. are still important, and SSD are still expensive.

Those people you refer to probably lack rationale then. you might have a point in say, Windows-based embedded systems, where yes, the exclusion of Luna might be a good idea given limited storage and memory space.

"Windows enthusiasts" are looking at ROS in hope to get compatible alternative, not a free drop-in replacement to save some money.

Your "enthusiasts" are probably looking for compatibility as well, hence the need for drop-in level compatibility. If you really want to break away from Microsoft's approach to OSes in general, just use Wine. No need for Windows-based drivers, and Windows services could probably be supplanted by their Linux daemon-process counterparts, and sooner or later somebody would probably have support for the former either as part of Wine or as an independent daemon wrapper around services.

[hto]

When all the sources are accessible, it should not be very difficult to cut away the parts, which somebody don't need.

[alexei]

It may be no contradiction between "minimization" and "compatibility"
From http://blogs.zdnet.com/microsoft/?p=842
While the Windows team has been working for years on reducing the dependencies in Windows which have made the operating system increasingly bloated and difficult to maintain and upgrade, it’s only been recently that the team has been able to create a separate, usuable new core.
Going forward, MinWin will be at the heart of future versions of Windows Media Center, Windows Server, embedded Windows products and more

ROS should try not to create dependencies to avoid "working for years on reducing" them

[RaptorEmperor]

I consider to be highly important to make the OS core not depent on all that "fancy stuff" that came aftre NT4 (I'm not talking about USB support and other new hardware, though). Unfortunately, along with improving user interface MS was adding non-removable "features" to each subsequent version of windows making them worse in each release.

He has a point here. I installed WIndows Vista Ultimate on a 20 gigabyte hard drive and it gobbled it all up because Windows wouldn't let me uninstall Media Center, Movie Maker, and a whole bunch of other non-essential Microsoft programs which don't even come with the cheaper versions of Vista. The most it lets you do is hide them from view. Removing, say, Movie Maker doesn't break compatibility, plus it lightens the load. I don't think he's necessarily talking about breaking compatibility, I think he's just saying that we need to keep ReactOS a lightweight operating system.

The project is pretty lean at this point, though, so you (Alexei) shouldn't worry too much about the size issue. Generally the nonessential junk is relegated to the "Download!" feature of ReactOS.

[alexei]

When all the sources are accessible, it should not be very difficult to cut away the parts, which somebody don't need.

Unfortunately, it is very difficult. With this topic I'm asking ROS team not to repeat MS mistakes and do things right from the very beginning. Also read this http://www.thehotfix.net/index.php/comp ... nd-for-all

[Z98]

To be perfectly frank, what I'm seeing is a general lack of understanding regarding the overall design and architecture of the Windows NT family. You keep going on about internal dependencies and asking us to reduce them without ever explaining exactly what internal dependencies you find problematic. All of the developers are fully aware of what MS has done since NT6 and continue to do in NT6.1. We already have an established code base, so whatever "mistakes" there were in the design are already there.

First and foremost, changing motherboards on Windows installs is possible. The limitation is a software lockin that ties to registration, not to any weaknesses in the OS itself. Second, allowing users to "customize" their OS requires a certain level of technical competence. No matter how easy we make it, unless you actually know what you're doing, you're going to screw yourself one way or another. Finally, again, do not presume that the reasons you are interested in ROS are the same as other community members or the developers. Our priorities may be nothing alike, and we will not compromise them simply to satisfy yours.

[alexei]

First and foremost, changing motherboards on Windows installs is possible. The limitation is a software lockin that ties to registration, not to any weaknesses in the OS itself.

Changing MB creates problems, such as http://support.microsoft.com/kb/314082 It's not because of OS "weaknesses", but by design to inforce the licensing that links software to the particular hardware.

Second, allowing users to "customize" their OS requires a certain level of technical competence. No matter how easy we make it, unless you actually know what you're doing, you're going to screw yourself one way or another.

It may be several "pre-defined" versions for those who don't know, but many people do know what they're doing. Why not to have "advanced" features both pre-setup like http://nliteos.com/ and post-setup like http://www.litepc.com/

Finally, again, do not presume that the reasons you are interested in ROS are the same as other community members or the developers. Our priorities may be nothing alike, and we will not compromise them simply to satisfy yours.

Of course, I can not prevent you from making ROS monolitic and non-manageable. It looks like I have to wait till you follow todays MS steps in componentizing their OS. BTW, componentizing, customization, and MB portability are not my personal requirements, but necessity well presented on numerous web forums.

Anyway, making ROS compatible with different versions of MS WIn, kind of "one size fits all" requires solutions. What about drivers? Let's say somebody has both old (W2K driver only) and new (vista driver only) hardware. What ROS can do about it?

[Z98]

I swear, you sometimes don't bother reading what I say. If you believe that ROS is somehow "monolithic and nonmanageable" then prove it. Go through the code and point out all the characteristics that make it so. Otherwise your assertions are pretty much meaningless since no one has any idea what you're talking about because you're being so vague.

You also have an odd notion of what being compatible means. Achieving compatibility is strictly a two layer proposition right now, at the kernel and application level. Drivers are kernel level compatibility for the most part, minus the usermode framework that's slowly gaining, and we've already stated we aim for 2003 compatibility there. To try and use a driver written/compiled for a different kernel version is a "you're on your own" type of situation.

Application compatibility is another matter entirely and it's one that MS has invested a great deal of work in. Solving it is near impossible, but their solution DOES work for the most part. Whether we adopt it, and it's very likely we will, is another matter entirely.

[Lone_Rifle]

I swear, you sometimes don't bother reading what I say. If you believe that ROS is somehow "monolithic and nonmanageable" then prove it. Go through the code and point out all the characteristics that make it so. Otherwise your assertions are pretty much meaningless since no one has any idea what you're talking about because you're being so vague.

You also have an odd notion of what being compatible means. Achieving compatibility is strictly a two layer proposition right now, at the kernel and application level. Drivers are kernel level compatibility for the most part, minus the usermode framework that's slowly gaining, and we've already stated we aim for 2003 compatibility there. To try and use a driver written/compiled for a different kernel version is a "you're on your own" type of situation.

Application compatibility is another matter entirely and it's one that MS has invested a great deal of work in. Solving it is near impossible, but their solution DOES work for the most part. Whether we adopt it, and it's very likely we will, is another matter entirely.

Actually, why do you bother talking to him. Just lock the topic and ban him for being reactionary or something.

[alexei]

Moderators, please remind Lone_Rifle to comply with the "Code of Conduct" http://www.reactos.org/forum/viewtopic.php?f=2&t=3709

Achieving compatibility is strictly a two layer proposition right now, at the kernel and application level. Drivers are kernel level compatibility for the most part, minus the usermode framework that's slowly gaining, and we've already stated we aim for 2003 compatibility there.

Thanks, it's clear with drivers now, but wouldn't it be benefitial to split application layer in two? I mean "OS management and limited application support" (take a look at BartPE) and "extended functionality fully compatible with MS Windows".

If you believe that ROS is somehow "monolithic and nonmanageable" then prove it. Go through the code and point out all the characteristics that make it so.

Not a long time ago MS proudly presented its MinWin. Can you show "ROS MinWin"? What about MinWin with OS management GUI to make MS jealous? BTW, I read MS used special tools to discover dependencies in their creations.
If MS Windows is "monolithic and nonmanageable" to users (I hope you're not going to argue with that) and ROS targets 100% compatibility then...

Read this Microsoft agrees: Windows is a "really large bloated operating system" http://apcmag.com/microsoft_agrees_wind ... system.htm Do you agree?

I'm kind of puzzled with your position regarding features MS does not provide for obvious business reasons and features that MS provides, but at the price of having permanently valnurable OS. In eight years MS could fix all holes in Win 2K, but they prefer adding new features to attract buyers. Are you going to implement all those security holes to be 100% compatible? Do you visit Black Hat https://www.blackhat.com to look at true picture? I'm not trying to offend you, just asking what you gonna do about that.

To me it's obvious that in modern world OS should have very small very secure core and minimal set of components for the particular need to minimize its valnurability footprint.

MS Automatic Updates is just admitting of being incapable of doing their job, though I've read about MS being kind of intentionally neglectant to fixing bugs to use them for computer penetration at government requests

Finally, what can ROS offer its users other then being free? I really hope it may do better then that.

[Haos]

Moderators, please remind Lone_Rifle to comply with the "Code of Conduct" http://www.reactos.org/forum/viewtopic.php?f=2&t=3709

Lone_Rifle doesnt need such remainder, you do.

Not a long time ago MS proudly presented its MinWin. Can you show "ROS MinWin"?

This is not our goal.

What about MinWin with OS management GUI to make MS jealous?

Again, we are not cloning MinWin, which is just a research project, not an OS for public use. No gui in MinWin... guess why?

If MS Windows is "monolithic and nonmanageable" to users (I hope you're not going to argue with that) and ROS targets 100% compatibility then...

Then what? You paired two unrelated statement and seem to push for any logical connection between them. This is called demagogy. In simpler words - bullshit.

Read this Microsoft agrees: Windows is a "really large bloated operating system"

You have obviously missed that part:
"A lot of people think of Windows as this really large bloated operating system, and that may be a fair characterisation, I have to admit" Taut said. "It is large, it contains a lot of stuff, but at its core, the kernel and the components that make up the very core of the operating system are pretty streamlined."

I'm kind of puzzled with your position regarding features MS does not provide for obvious business reasons and features that MS provides, but at the price of having permanently valnurable OS. In eight years MS could fix all holes in Win 2K, but they prefer adding new features to attract buyers.

Typical slashdot crap.

Are you going to implement all those security holes to be 100% compatible?

In what way those security holes are required for compatibility?

Do you visit Black Hat https://www.blackhat.com to look at true picture? I'm not trying to offend you, just asking what you gonna do about that.

Why should we? Are we perhaps fixing Windows source?

To me it's obvious that in modern world OS should have very small very secure core and minimal set of components for the particular need to minimize its valnurability footprint.

Obvious doesnt necessarily mean true.

MS Automatic Updates is just admitting of being incapable of doing their job,

It works just fine.

though I've read about MS being kind of intentionally neglectant to fixing bugs to use them for computer penetration at government requests

Do you read about UFO abductions and Atlantis with the same seriousness? Its around the same level...

Finally, what can ROS offer its users other then being free? I really hope it may do better then that.

It can offer them pain, hard work, blood and tears. At least its my case. Also, it can be a source of knowledge on NT architecture, but reading your posts i have a feeling its something you dont really care about.

To sum up. Your knowlegde about NT is zero. Your knowledge about MS Windows is zero. Your knowledge about ReactOS project is also zero. All your information are based on slashdot and similar sites, that are known for they lack of understanding on the afforementioned topics. From this toxic mix you draw conclusions that show your total ignorance on the subject. The MinWin part proves it just fine... I can only suggest that you stop reading crapm, ranty articles and start digging into some reliable sources. Then get back and we can discuss.