Slashdot Mirror
ReactOS Revealed
reactosfanboy writes "DRM Hacker Alex Ionescu explained the internals of ReactOS in a recent talk. Ionescu indicates that ReactOS is nearly 100% binary and API compatible with the Windows 2003 kernel, and that they are aiming for full Vista compatibility. Ionescu attempted to demonstrate ReactOS but only succeeded in installing it after two BSoDs. This alone should make it clear that ReactOS is still not ready for prime time." In what may be a red flag for Microsoft's lawyers, ReactOS is described as "an environment identical to Windows, both visually and internally." Here are slides from Ionescu's talk (PDF), which might prove more useful than the video offered in various forms at over 450 MB.
How about a link to ReactOS in the summary?
^^
There was a minor scandal last year when the ReactOS project had to be suspended after it was found that some developers had been exposed to the real Windows source code (which is available under shared source licenses in some contexts), and after a thorough audit the project continued.
The ReactOS people are taking the risk of copyright infringements very seriously.
You are not alone. This is not normal. None of this is normal.
So you won't have to actually run Windows in order to run Windows programs such as Photoshop, AutoCAD, and most video games. WINE isn't good enough for everyone.
"In what may be a red flag for Microsoft's lawyers, ReactOS is described as "an environment identical to Windows, both visually and internally."
Oh, please... While I have no doubts MS will try to destroy ReactOs when it becomes too popular, the developers have made painstakingly difficult steps to ensure the proper reverse engineering is done ( http://en.wikipedia.org/wiki/Black_box_testing ). They can sue all they want, they can't win this. (They can however make it an expensive legal wrangling...but then again, since it's open source, it's difficult to imagine any single lawsuit will be able to end the project).
--- "To pee or not to pee, that is the question." ---
But you can see them here: http://www.alex-ionescu.com.nyud.net:8080/wloo-tal k.pdf
The zipped vmware image is 19 megs. So I'd say its a bit lighter ..
The incident had nothing to do with Windows source and it was certainly not minor. It was due to certain parts having been implemented by the same programmer that had reverse-engineered them and was consequently "tainted" if the project was to adhere to its principle of black box testing only. That programmer was the very same Ionescu as here. The result of the unfortunate incident was that the programmer that found out about it (Hartmut IIRC) resigned from the project and the audit that is still going on was started. I read some of the discussion about it on the mailing list and apparently there was a great deal of concern about Ionescu's contributions since they came too quickly and were too good to be the result of just blax box testing (but not all is available for everyone so there could've been something else as well that resulted in the conflict between programmers - the whole project was to some extent in jeopardy, though). The only thing you're right about is that they take copyright infringement seriously but that has nothing to do with that incident.
hey, that sounds mighty familiar...
Actually what happened was someone alledged that someone on the development team had decompiled actual MS Windows dll and binary files, and included the resulting code into the ReactOS codebase.
Unfortunately for the most part ReactOS is just wine that self boots. They borrow heavily from the wine codebase for their win32 api compatibility. Naturally the driver compatibility doesn't borrow from the wine codebase. The real point of ReactOS isn't software compatibility but environment compatibility. The goal is to be able to install all your windows apps on a machine that only supports windows (due to poor ACPI support in linux or poor video driver support or any of a number of incompatibilities). However windows application compatibility is rarely if ever better than wine with reactos.
Application Compatibility - Wine can never offer as much compatibility as ReactOS. Since ReactOS actually shares Wine's code base, it is highly unlikely that ReactOS will have significantly better compatibility than Wine.
Not true. ReactOS uses some of Wine's codebase, but many key areas (window management, memory management, thread support, etc.) are rewritten from scratch to be more compatible with Windows own implementations of these. I've seen Wine fail to run applications before because of some subtle difference in how Linux handles these tasks to Windows. ReactOS can eliminate these issues.
Driver Compatibility - ReactOS can use native Windows drivers. Projects like NdisWrapper have shown that it is possible to use Windows drivers on Linux too, if enough people are interested.
Yes. Unfortunately NdisWrapper doesn't really work very well (my limited experiments suggest only about half of the cards out there work with it), CaptiveFS is slow, and no other projects have produced useful results in this field. This is because running a Windows device driver without a Windows kernel is quite tricky. NDIS drivers are a simpler problem: NDIS was originally developed to be an open, cross-platform specification by 3Com. MS have embraced & extended it since then, but at its heart it is still much more portable than many other driver types used by Windows.
There is also the huge issue of using binary drivers in an open source kernel. It still hasn't been settled whether or not this is ultimately a good or bad thing. However, it is generally accepted that open source drivers are much better than binary, and ReactOS would provide absolutely no motivation for hardware vendors to ever open their drivers, or even to target ReactOS as a platform.
Frankly, there are a lot of us who have become fed up waiting for working open source drivers for our hardware and would rather just plug a black box in and be done with it.
Also, if one desired you could rework something like KDE to be VERY similar to Windows, I believe that there are already distro's who try to do this (such as Linspire). There are still differences, but not really significantly more than between Win 98 & XP.
Actually, there are very substantial differences that can be deeply annoying because they're about the way the basic system works. Details like which control panel applet you use to start or stop services (e.g.) aren't as annoying (to me) as the lack of feedback when a program is starting (KDE does have some feedback, but it doesn't show if the program wasn't started from the window manager, whereas Windows will show it however you start your program). This can't be fixed easily in Linux: it needs the kernel to provide feedback to the window manager to inform it when a graphical subsystem program is in the process of starting up. Linux doesn't have such a thing as a graphical subsystem program, and the window manager is not a special process that could easily receive such feedback from the kernel.
I think a lot of people have missed the real point of ReactOS. Including the developers, by all appearances. Windows won't run under Xen paravirtualization. There's no reason ReactOS couldn't be ported to it, however.