Steven Edwards On The Future Of ReactOS And Wine

Alex_Ionescu writes "WineHQ brings us the scoop on the latest developments in ReactOS, as well as on Steven Edward's excellent job on porting Wine to MingGW and linking the two platforms together. This is an interesting insight into the WINE and ReactOS project, and a must-read for anyone interested into the future of Windows-replacement projects like these."

Comments

Why clone Unix?

[jschottm]

Unix was cloned for a number of reasons. First, it's (or very similar clones) been used extensively in the teaching of college level OS courses (Lyons, Tanenbaum, et al), so when students decided to write their own, it was natural to model their OSes after what they'd learned. As has been beaten to death in the past few weeks, Minix was specifically designed to be small and compact so that students could understand it in a semester. How many versions of Windows is that true for? *BSD obviously came out of the education system, and Linux was written in response to Minix.

Second, Unix systems have been established a track record of power and reliability (yes, there have been very bad Unixes, and they tend to have been removed from the marketplace). Windows ... hasn't. It's gotten remarkably better, and a good deal of its problems are due to 3rd party drivers, but my well maintained W2K desktop and XP laptop still need to be rebooted every two or three weeks. And there's the never ending string of serious vulnerabilities. At an OS level, Windows has a lot of nice ideas. The problem is, most discussions about them seem to run, "They had a nice idea, but..."

Windows is changing rapidly, in ways that are likely to make programs incompatable with older versions (the better to force upgrades with, I'm sure). I'm sorry, but if after 7 years of work the project is almost within grasp of being able to use a DHCP client, I don't see any way they can keep up with Microsoft. If they want to work on it as a hobby and have fun doing so, more power to them. I just don't see it as being something overly useful. Screenshots of minesweeper (with poor graphics) aren't what I want to see. I want to see a version of Group Policies, Active Directory capability, and so on.

*BSD and Linux suceeded, not simply because of price, but because they were *better* in various ways than the competition. Microsoft has a tremendous software and driver collection, and has begun to do some really cool stuff. OS X has a simple UI that many people adore. What does ReactOS bring to the table, if it's three generations behind Microsoft? DR-DOS was cheaper and better (IMO) than MS-DOS, but Microsoft still ground it underneath their boots.

Re:Why clone Unix?

[Mycroft_VIII]

Well from reading the article (please don't kick me off, I won't let happen again honest:) )
I would say two things
1) an open source code.
2) as part of thier developement process thier creating usefull tools and info to help developers of windows software port to linux, or even write code that easily ports from one to the other.
The reason Dr. Dos and other failed was in part do to the fact that it depended on income to succed and thus could go belly up financially. Much harder for an open source project do that.
Also Microsoft wrote code in thier apps that generated false error messages in some dos replacements, giving the false illusion that the dos replacement was buggy or incompatable. FUD wasn't used because microsoft was feeling sadistic, it was used because it worked.
Of course one of my favorite reasons for writing any open source/freeware code is 'why not?'

Mycroft

Re:Why clone Unix?

[burns210]

after 7 years, the infrastructure is almost there to use a dhcp client... but dhcp, is a layer 7 protocol, and that implies a LOT of stuff working below(in ip, tcp, etc...) and a lot of the OS working, for something as high-level as dhcp to work. if it was ping, then big deal, but dhcp implies a lot. Besides, the firt 60% is the hardest, just plain work that isn't fun... and it goes the slowest, but once that 60%, or whatever the magic tipping weight is, a lot of stuff just falls into place... it might take 10 years to achieve X, but once X works, that means Y and Z both work, and Z and X combine to make A, B and C all work...

it snowballs my friend. It is a slow rolling snowball, but it grows and grows and grows.

Re:Why clone Unix?

[jschottm]

1. True. I find it more academically interesting that other groups are using/considering using the kernel with a different user environment than the concept of cloning Win NT. But I can't help but wonder if the effort would be better spent working on Wine or allowing Linux or FreeBSD to use Windows drivers. But as you said, why not? It's their time and their choice.

2. Open source may not need money to survive, but it does require "the itch" that causes the programmers to scratch. Remember, writing the software is the easy part. Testing the code and making it bulletproof is the hard part. I've seen many open source projects get 90% done and fall apart because they find that there just isn't that much consumer interest in it, and there wasn't the motivation to get over the hump of sitting there are 2 in the morning trying to track down that memory leak that seems to only happen every 27th time a function is called.

As far as DR-DOS, what makes you think that if ReactOS were to actually threaten MS, what makes you think they wouldn't squash them through use of FUD, patents, or other such measures? Given that MS Office is still 'the killer app' for offices, I doubt it would take much code to make it develop mysterious errors when running under ReactOS.

Other than the fun-for-the-programmers aspect, I'm still just not seeing the target market for this. With home users, either they need the absolute basic stuff (word processor, e-mail, spreadsheet, browser). I suspect that by the time that ReactOS is finished and stable, there'll be cheap Walmart or AOL branded Linux boxes that fill that role nicely. Home users looking to play games won't be interested, because modern games will be so far above what NT can handle. (I suspect that MS has more people working on DirectX alone than the entire ReactOS team.) And most business users will shy away from anything that's not heavily tested. My employer provides me with a copy of VMWare and WinXP, because it's far cheaper to purchase those than to have me burn hours futzing around with a first generation cloned OS.

So we're back to "why not," which I think is a better answer than, "why clone Unix?"

Re:Why clone Unix?

[JohnFred]

This is just NOT true. It's revisionist history. However, the nice thing about the web, is that the original article written by the first person outside MS to discover the code is still online.

This is what's known by historians as a primary source.

Re:Why clone Unix?

[Olivier Galibert]

DHCP only implies UDP, and that's rather easy. In fact, I participated in a project where it took ~6 months to get IP/UDP/ARP/DHCP running in VHDL on a FPGA, and let me tell you, it's way harder than in pure software.

TCP on the other hand is really complex.

OG.

As cool as the concept of ReactOS is...

[Quattro Vezina]

...I can't help but worry that Microsoft is going to screw them over. I think that the more ReactOS develops, the more likely we'll see an immoral patent-infringement lawsuit from Microsoft against ReactOS.

Thanks...

[NamShubCMX]

Although I don't see the use now, I know that 10 years from now on, I will say THANKS too all the developers that will have allowed anybody to use their old unsupported softwares...

Good progress

[WhoDaresWins]

I've been surprised how far ReactOS has come along. I didn't expect them to progress this much by now. I could actually install the last release on real hardware and it installed and ran AbiWord just fine! BTW a lot of people seem to have problems with the install CD .iso based installation of ReactOS. There is a simpler way to run it if you have a FAT16 or FAT32 C:\ boot partition, just download the binaries and unzip them to C:\ReactOS\. Then just boot from a DOS floppy and run aboot.bat within C:\ReactOS. Works like a charm everytime (for the past half dozen releases anyway). BTW if you insatall the VESA mode VBE driver (search the kernel mailing list) then you can get AbiWord working in true color. Its impressive to see it working considering how far ReactOS has yet to go.

Re:Good progress

[gronnsak]

Then just boot from a DOS floppy and run aboot.bat within C:\ReactOS.

A Canadian project, is it?

A perfect name

[foreverdisillusioned]

porting Wine to MingGW

Hey, why not call it that? "Port"! I dunno if renaming a port is unusual, but if so I think we can make an exception in this case.

MingGW?

[achurch]

Would that be the Chinese version?

Why?

[AmVidia HQ]

I can see the usefulness of Wine, in running legacy programs as well as serving as a bridge between Windows apps and Linux. But why write an entire new OS for this same purpose? I just don't see the point of re-inventing yet another Windows wheel.

Perhaps starting from scratch (ReactOS) is easier than the writing the middleman layer (Wine), which is still playing catch up after many years?

(Any flames was unintentional. I would love for either project to succeed, I just want to know their merits)

Re:Why?

[KJKHyperion]

You look at Windows and you think "past" and "legacy". We (ReactOS) look at Windows and think "future" and "innovation"

This you say is (my personal opinion, not the project's official position), a lot like people deluding themselves in thinking that a Linux kernel for the next Windows would mean instant improvement. If you knew a bit more about Windows you'd knew it doesn't even make sense - the Windows kernel has proved to be a sound design and I'd be happy if we could at least duplicate it. It's the system services that need a serious redesign, dragging the legacy microkernel corpse as they are, and I'm already doing some research in that direction (currently reimplementing the console model to be driver-based instead of server-based, and allowing custom UI implementations. This will give you, the user, custom terminal emulators, a more efficient SSH server and a job-controlling shell - even a port of your beloved GNU screen. Next stop: overhauling the service manager and rationalizing the user-mode startup sequence)

And please realize that a supposed intrinsic and purely technical superiority of Linux exists almost entirely in your mind - old kernels (2.4 and earlier) weren't that hot, and when you say "Linux" you generally mean "GNU" (those who say "Linux" meaning "Beowulf" or "OpenMOSIX" are the minority) - and realize that "commodity" also means "largely irrelevant" and "expendable" - when ReactOS gets good enough, it could replace even Linux for some people

Probably will hit 1.0 a year after Duke Nukem

[jmorris42]

Ambitious, but not likely to be relevant.

Wine is almost 10 years old and yet to ship a 1.0. And already bitrotting away because parts are still win16 (from reading the article) because they were coded pre windows95.

DOSEMU did eventually ship their 1.0 version... and was promptly deleted from the RH disks in the next rev as obsolete. It 'succeeded' because they were cloning a dead OS that didn't keep changing. If you count success as finishing long after it would have been widely useful.

Now we have ReactOS cloning Windows NT4. And will perhaps get it 90% feature complete in another few years. And then spend the next half decade completing the remaining 10% by which time NT4 will be so obsolete nobody will care. Of course they are already trying to shift their target to NT5.1 (XP) but like Wine, they just can't code as fast as the infinite monkeys at Microsoft.

As for their retort of "Why clone UNIX?" I have an easy answer. Because it is USEFUL. Microsoft's stuff isn't worth cloning and by the time a clone is finished they will have either won, forcing everyone into a DRMed hell where only their signed OSes even boot or we will have made them moot.

Re:Probably will hit 1.0 a year after Duke Nukem

[jmorris42]

> I mean, who cares who writes the drivers, or whether they even
> get written?

Kinda hard to boot without the basic device drivers. So yes, it is important unless the plan is to leach off of the Windows driver set and that really isn't the GNU way.

> Free software is about freedom.

Agreed. But just how much Freedom does one expect to find chasing Microsoft's tail lights?

> to OSIers, it's all about how useful it is...

Well it does need to boot and run programs ya know. That sort of fundamental functionality is what I'm questioning the ability to create in a useful timeframe. If DOSEMU and Wine are a guide, ReactOS will be more of a MAME/MESS sort of nostalgia trip. Don't see how that advances the cause of OS/FS other than increasing the skills of the developers.

> Maybe you're a new Linux user, or maybe you aren't one,

Examine the URL at the top of my posts. WIth a shrinking list of exceptions, my current working set of software would pass the RMSLint test.

> It was a pain in the ass, but we did it,

Hey, the first time I saw a blurb in Byte about Linux I had a boner for it. A year or so later I actually found a boot/root disk of .02 or something like that on a BBS and tried it. It wouldn't see my HDD and I was too poor at the time to go buy new hardware so I put it aside. Finally in 1993 I found the Yggdrasil Linux/GNU/X CD and after hacking around and doing much research managed to get it to see my funky Tandy Outlet store scrounged 1X CD drive. (Borrowed a different drive to initially install from, had to compile a patched kernel on a 386sx-16 so don't even try talking to me about pains in the ass because I have the t-shirt.) It was another year and more hardware upgrades before I reached a point where I could start planning to move most of my day to day operations over.

The point is that even ten years ago, a two column inch mention of serious work on a Free OS was more than enough to get me interested, but I just don't feel any such burning need for ReactOS. And I tend to doubt many others will either.

Linux was possible because of the decades of UNIX tradition, a published POSIX standard to write to, almost a decade of GNU's work before Linus ever wrote version 0.01, the X Consortium's codebase, etc. Where is the great body of enabling work that ReactOS is to draw from? How many Free Software types care about Win32 enough to write code for it? Hard enough to even get most to support Win32 as a port because Win32 is fugly and makes for messy code.

This poor suffering disucssion

[ChiralSoftware]

People on /. love to talk about all things to do with MS vs. Linux and yet this poor suffering discussion only generates a couple dozen comments. I will add in something about the best real-world way to use Wine: Codeweaver's Crossover Office. It really works. Microsoft Office 2000 works perfectly in Crossover Office, and Office XP is completely usable for day-to-day work. It is amazing how well these things work and how well they integrate into the Linux desktop. Wine is the foundation of it, and it works. If it works this well now, how will it be by the end of the year? They also have a cascade effect, in that if they solve a bug in MS Office, that might also solve bugs in many other untested applications. I have noticed that the unsupported apps work better and better. Wine is a relevant and cool project. The thing that might make it irrelevant is that Linux desktop software options are catching up. I think that OpenOffice.org is already better than MS Office in many respects.

-----------
WAP software

Windows drivers to linux conversion

[StripedCow]

Would it be possible to use the Wine code to run hardware driver code written specifically for windows, under linux? I guess it would be nice if this could be done without having the complete "wine" emulator in core, only the necessary components. Perhaps someone could write a windows driver -> linux driver converter, which takes the windows driver object code, and links in the necessary components from wine (only those win32 functions called by the driver, plus dependencies).

I see where ReactOS could be REALLY useful

[nickol]

It's embedded systems. Today, Microsoft has Embedded NT 4.0 which is too expensive, WinXP that is too heavy for most embedded computers, and WinCE embedded which is a kind of a joke.

What we really need for those PC104 and other small boards is an OS with the following features:

open-source and configurable

reliable and stable

small resources requirements

working from ROM

Win32 compatible, supporting DCOM and MS-style networking.

There is no need for DirectX, scanner support and such. It looks much like that despite Microsoft declares embedded systems support as one of their primary goals, they just do not know what to do.
WinCE is for PDAs, not for industrial systems.

Re:I see where ReactOS could be REALLY useful

[Teancum]

I think you miss the point. By using ReactOS, you cut out the software bloat that MS throws in, and you can pare down the kernel to the bare elements. ReactOS allows you to work at a command-prompt only level, so if you have an application like a server or embedded controller that doesn't need a GUI, ReactOS is perfect. While I don't think this is a deliberate goal of this project, it would not be all that difficult to make a single floppy version of ReactOS, like QNX has done.

The really nice thing about this is if you have components you are adding to an embedded system, and the manufacturer has Windows device drivers, ReactOS will recognize them and they don't even need to be recompiled. This opens up a whole range of equipment options that would not normally be available under embedded Linux (although most embedded equipment companies are supporting Linux now... this wasn't always the case).

I don't know what is cheaper than free software, and the point here is that this increases flexability. I don't know about what you mean with hardware requirement, but I don't see too many 386 CPUs anymore, even among embedded systems. ReactOS will run on most of the common CPU systems found in embedded systems, particularly if you are already looking at a Linux-based system as well. This is an option, not a requirement.

If you are talking about using the Microsoft version of the NT 4.0 kernel in an embedded system, I would have to totally agree it is a mistake. I would still be careful about the "uneducated developers" you are thowing stones at, because there is a huge difference between the NT kernel and Windows CE (which truly is for the clueless developers). On raw technical merits, I would stand behind an NT-based kernel as much if not more than a unix-based kernel (like Linux).

Re:What is ..

[Lussarn]

It's a fast cpu-emulator with JIT compilation and such. About 100 times faster than bochs.

QEMU

Re:Usefulness of PPP

[Mycroft_VIII]

Most of the world is on dialup. This includes alot of hackers. and some opensource contributors are kinda poor and don't have home networks, Mine is Two computers right now. and I'm stuck on dialup. and other that briefly using slip (early 95 iirc, only 1 isp in town at the time and thats all they used first 3 months) I've been on nothing but ppp.
Oh and you don't have to be 'leet' to code, even on the big opensource projects. You just have to be able to code. That's the strength of open source anyone can spot somthing, anyone can write an answer. True 'leetness' (skill talent and experience) increase your versitility and odds of comming up with somthing really cool, but even the HURD and GCC and Linux need that four line patch to swap some bytes around on some obscure file format or device driver.
Don't buy into myth of 'leet' or 'real' hackers as all fitting some stereotype or other, that's hollywood talking. They fall into the same general mix of poor and rich and so on as anyone else.

Mycroft

WineHQ screen shots

[linebackn]

Since we are mostly the topic of Wine, on the WineHQ screen shots page http://www.winehq.com/site?ss=1 they have an old link to my site ("Nathan Lineback's Wine Screen Shots") If they want to keep linking to it, It needs to be updated to point to http://toastytech.com/guis/wine.html The current link points to my very old pla-netx address that stopped redirecting to my new site recently.

I had tried e-mailing some contacts listed on the site, but there has been no responce. Who should I contact? Thanks.