I was under the impression that a system would only use virtual memory if there was no space left in the RAM. Because virtual memory is stored in a paging file on the hard drive, it's much slower than actual RAM, and will increase hard drive access times causing potential slow downs but I always thought that the slow downs will only occur if the app is making heavy use of virtual memory, not actually virtual memory itself. If you had enough RAM and virtual memory was enabled, I was always under the impression the system wouldn't access that paging file at all. Since upgrading the RAM in my laptop, I've turned off virtual memory and haven't had any problems so far with any game.

Bringing up the windows services was a good point, back then I didn't really check which processes I needed and which I didn't. I'm sure I could have freed some some memory (and made less use of virtual memory) if I did that.

Hmm, well, reading the game page, I thought that maybe Bohemia were pulling OW along with the Arma series because of the multiplayer component not working, and that maybe it was up and running on Steam. I thought to myself, perhaps they want all of their games sold with full capabilities.

That apparently isn't the case though, going by what you said. Multiplayer is functional and it's still DRM-free, yet they're removing it from the GOG store for some reason... :/

Okay, I know that ORIGINAL WAR was not on sale, but I'd
have no problem with it being listed in the sale...

If I had not read the notice on the bottom, I would have
not gotten the game.

Now, I could have survived without it...but it would have been dull.

I wasn't planning on buying these games but the fact that they're going off the site made me make an immediate purchase.

If you have 256Mb ram, Windows will by default add 512Mb of Virtual Memory, meaning it will SAY you HAVE 768Mb total RAM. Once the memory is allocated then it works with it.

Actual Virtual Memory will try and swap out inactive processes first so active processes can work (So if you have a PDF file open that takes 32Megs to read and you aren't using it but instead using Office software, the inactive data is stored in a pagefile unless it's required later). When used that way it's fine; However if you have in the same scenario a program that requires 500Mb or more (games, cad software, 3D Modeling for example) and you only have 256Mb, then there is constant page swapping which will slow the system to a crawl.

Some API's will memory map a file to memory without actually loading it and only load it when needed, while others using classes and C++ make huge amounts of objects and it's speed is not a problem when there's no real delay for memory access but the VM access will make it unplayable, as it not only needs to access, but create and then free those items later.

I'm actually really really surprised M$ didn't do similar to Linux with where a pagefile is created in memory but it uses compression thereby dropping page access and using compression/decompression instead (<i>and with like [url=http://en.wikipedia.org/wiki/Lzop]<span class="bold">LZOP</span> it's instantaneous, or Zlib is fast and decent), and offer real pagefiles only when ZRam couldn't handle it.

Course M$ also doesn't allow us to create or use our own filesystems, so something like SquashFS isn't avaliable, which is a big bummer because I love that FS.

Yes and no. In XP yes you probably could have (as it is sorta a variant of NT), but with the updates requiring ALL drivers to be signed in order to be used, you can't just compile or install any drivers without disabling those features. Disabling them also is a pain from what i can tell, and it doesn't guarantee you can use them afterwards. And to sign anything you have to pay a license fee. Some software and drivers have been signed which resolves installing and using them like imdisk.

As for ZFS, i believe there are many people who want to use it (I'm curious too), however the license and copyright behind it reserves it to Sun Microsystems OS's only, or is incompatible with the GPL license since open source does not share freedoms including modifying source, distributing it, enhancing it, etc. There are free attempted non kernel port for Linux systems, but it requires a lot of workarounds (in the form of a userlevel driver).

In the same way with signed drivers and licensing, M$ isn't going to write the drivers or support them as it doesn't help them. Far more likely it's a can of worms M$ won't open.

edit:
from the site: The Windows 2000 IFS kit is obsolete

Glancing it over the driver kit available is for windows 8.1 and i won't ever touch windows 8. Yeah..... fun...

Disabling driver signature verification in Windows 8 is easy; it just takes a reboot. (It's buried in advanced options in the Control Panel, since it's not something most people will ever need to do.)

In Windows 7, it's a bit more techy but not hard; two command line statements. I've run my Win7 system that way basically since I got it with no issues; my sound card didn't have a signed driver forever.

It's a similar (but slightly different) command line for Vista.

They have completely removed all games off GamersGate except Carrier Command and the Google's Cache is May 30th to June 2nd. So it looks like its completely coincided with GameSpy's shutdown.

They may have assumed Original War required GameSpy or thought it was best to completely pull out of GOG. I don't know. I have sent Bohemia an e-mail asking why they are removing it.

Easy or not it's an annoyance and although I understand it's a way M$ is trying to limit security issues, Windows still has (and will always have) the largest number of viruses, trojans, worms, and hacking vulnerabilities of any system, it's any wonder it's one of the most used. Many of the tools they've been removing are quite annoying and basic and needed, but aren't there so the standard stupid user doesn't try to use them. (Gee what does 'format' do?)

It's really not an easy OS to do extensions to the system, being proprietary and closed source you don't know what you're really working with. They also pushed towards using .NET which I suppose it good since I see some good things about it, but at the same time it's still very proprietary and anything you write there isn't going to be very portable, especially if it's like C#; No you're entering the same problem that was present in Java for a while where the company could enact terms for using their libraries at any time which makes anything you make pretty much non-portable and non-free.

But the conversation is drifting quite a bit regarding adding your own Filesystems. Quite simply, it's probably a bad idea and better to run it through emulation or not use Windows at all for that type of thing.

I'm very interested in the response (assuming you get one). If you wouldn't mind, please post it if/when you get it.

Thanks Stucuk.

The user has the choice of keeping the security signing provides or ditching it, and as abrigati pointed out, ditching it isn't difficult. Ideally, though, windows would allow this on a per-driver basis rather than only as a system-wide setting. (But there is tons of crap in Windows, Linux and every other OS I've used that is far from ideal.)

Yes, the separate free product "Windows 2000 IFS kit" is obsolete (and may or may not be available anywhere on the MS site anymore -- a lot of their older stuff is still available for download, but I'm not going to bother checking for this one because I don't actually want it.) The free "IFS Kit", however, is not obsolete, and as the page says "is distributed as part of the Windows Driver Kit (WDK)" (and has been since XP).

If you were actually serious about developing your own file system (rather than just bitching about it on the internet), you would notice the 8.1 driver kit works for windows 7, 8 and 8.1, and that you can still download the driver kit for XP (which also works for Vista) on this page.

Don't get me wrong. MS sucks. Windows sucks. (Linux sucks much more IMO.) But if you're going to point out suckage, try to hit the damn target.

You might want to check out something called "documentation". Windows development docs tend to be leagues ahead of Linux. (The argument that you can look at the source for Linux doesn't really work on me, because trying to figure out a zillion lines of code written by a thousand code monkeys doesn't really result in me "knowing what I'm really working with", not in any useful sense beyond reaffirming that raw code is no substitute for proper documentation.)

Trying and succeeding. They've basically wiped out an entire classes of malware -- those that hooked into the OS itself.

That's just a completely unsupported claim on your part. Note that # of successful attacks does not correlate to # of vulnerabilities and has much more to do with:

Specifically, it is by far the most used open OS for end-users (who don't take the kinds of security precautions that corporations tend to take with their servers, and end-users are statistically speaking far more susceptible to things like trojans because they don't have a clue when it comes to security). Note that I emphasize "open" (as in people can create/upload/download/install their own apps), because with a walled-off garden like iOS effective security becomes significantly easier. That of course just gets right back to driver signing -- by requiring driver signing MS has taken a large step towards wiping out driver based malware, but they've turned drivers into a bit of a walled-off garden (though not as bad as iOS -- your drivers don't need to be approved by MS or anyone else, you just need a certificate from someone whose chain goes up to the MS code signing authority and that is about verifying your identity). If you want them to eliminate more malware, a very practical approach (if undesirable) would be to require all executables to be signed (or worse yet, be like iOS and require review and approval by MS). (A very practical and desirable approach IMO would be to require code signing by default but allow the user to white-list specific executables/paths as not subject to the code signing requirement, and don't provide any programmatic way for non-system code to read or write the list of where the "unprotected places/executables" are.)

They're removing "quite annoying" tools? Sounds good to me. :-P

I'm using vista and it still has "format". Are you saying 7, 8, and/or 8.1 doesn't have it? I didn't know that, but it wouldn't really surprise me, nor annoy me. And it's hardly "basic" or "needed". I practically live in the command line (and emacs), and I don't think I've used the "format" command since Windows 98 (if then). Format is a rare operation for most people (even most devs) and the GUI tools MS provides for formatting disks work quite well. (They're certainly more friendly even for the tech-inclined than using fdisk and format.) And if you need to script it, I'm sure MS provides some way to do that -- you can use stuff like diskpart which is way better than fdisk ever was (e.g., with the built-in ability to change a partition's size without losing the data currently in the partition).

So now you're telling me that mono ("An open source, cross-platform, implementation of C# and the CLR that is binary compatible with Microsoft.NET") doesn't exist? (I'm starting to have some doubts about the reliability of the information you provide!)

I don't think Mono uses MS's libraries anyways. MS only provides their libraries with a "for debugging and reference purposes" license, so Mono has to have its own library implementations or it's already in violation of the MS license.

And from my perspective (and this is just my opinion, but it's the opinion of someone with 30+ years of development experience at every level of the digital stack from app creation down to soldering transistors together to writing my own programming languages and tools), VMs (and other portable application formats) are a damn good thing. It's about time we get away from being tied to a given hardware line (x86) and all of the cruft that has (and is still adding at an ever increasing ridiculous rate) and away from being tied to a given programming language. And there's not many of note -- there's Java (JVM) and there's .NET (CLR) and... I'm sure there's literally thousands of others but other than some that I've worked on myself, I really can't even think of any others. And between the CLR and the JVM, I believe the CLR is the better VM spec. (I'm not counting LLVM here because it isn't intended to enable end users to run portable applications, but rather is intended to allow developers to crank out a slew of platform-specific applications, which really sucks if you are using a new platform that never got its own version of the app. Plus LLVM only addresses CPU dependencies and not OS dependencies. Something like ANDF would be closer, but still doesn't allow avoiding OS dependencies.)

So yes, damn MS for creating the best widely used VM spec in existence. MS just sucks for doing that. And they're not just giving all their code away but only providing a free license for reference purposes? Damn them! (And if your posts aren't about bashing MS, then maybe you'd explain your constant use of "M$", especially when we are talking about a free driver dev kit and a freely provided VM spec and the code signing license fee in question doesn't even go to MS, it goes to whatever cert authority you pick.)

The open source community is free to define its own VM (or portable application format) spec. MS isn't stopping them. I'm not stopping them. (Lack of time, talent, vision and economic motivation may be stopping them though.) So far I haven't noticed anything but Mono (CLR clone), and I don't think there's anything wrong with having Mono around, but if the argument is that "MS VM sucks", then to back that up there needs to be something better, not just a clone.

"Quite simply ... bad idea" - so simple that you can't even justify that position.

BTW, I've actually gone through the process of writing a transactional log-based persistent object store (that's currently being used by a Fortune 500 corp). If I were to repeat a similar effort (we're talking minimum many months of full time dev work, a hundred thousand dollars worth of dev time) for a file system, I would not be remotely dissuaded by a measly $140 fee to get a 2 year code signing license and get the driver signed. If all that's keeping you and everyone else from using SquashFS on windows is that you can't chip in and come up with $140, then there simply is not any real demand for it. If ZFS came out tomorrow as a windows product costing $200, I would buy it (assuming reviews indicate the port is reliable and complete). I might even pay more than that for it. I'm already paying a lot more than that for many TB of storage so I can store data in quadruplicate (one original, one "on-line" (disk in machine) backup, two "off-line" (disks stored unconnected to avoid lightning/power-supply damage) backups).

awaiting post for edit since it's slow

edit: seems it won't let me post this. Fun.... If I can get it replaced later I will, but for now I'm tired of fighting with it.