StarDestroyer.Net BBS

phongn wrote:

Darth Wong wrote:UNIX had a proper multi-user layout long before then, on even shittier hardware. It was a short-sighted design decision.

Well, didn't UNIX get the luxury of hardware designed for multi-user operating systems? It would've been rather difficult to do any sort of multiuser work on an Intel 8080.

Windows never ran on an Intel 8080 as far as I know.

Darth Wong wrote:UNIX had a proper multi-user layout long before then, on even shittier hardware. It was a short-sighted design decision.

Unix had the luxury of being originally a time-sharing OS and in a time where CPU time costed non-trivial amounts. DOS/Windows originated on personal computers where there wasnt a cost for CPU time and there was only a single user.

That makes for dramatic differences in design decisions.

Darth Wong wrote:Windows never ran on an Intel 8080 as far as I know.

Old versions of MS-DOS did, and the intel x86 line could not do multi-user/pre-emptive multitasking until the 286 which was many versions of MS-DOS later and several versions of Windows 1.0->3.0, The 286 preemptive multitasking had such a huge preformance hit that it really wasnt until the 386 was around (and the increases in memory that occured by this time) that it was doable on a desktop computer.

By then it was too late, too much existing software was already locking in bad design choices.

MS has been locked into a backwards compatibility problem for a very long time. Too much stuff has relied upon the APIs remaining the same. Heck, there are some bugs that MS can't fix because programs rely upon the broken behavior, and some company, somewhere needs that program.

MS-DOS (and Windows, which ran as a UI over DOS) was designed for a single person microcomputer. Programming in multi-user functionality wasn't just not a laudable idea, but in fact an asinine idea. Unix was designed for a minicomputer. It had to be designed with multi-user capability from the start.

You're complaining that not only did the MS programmer not predict the future, but failed to do something that would have actively jeopardized the ability for that future to come about.

Oh, and MS-DOS started off on the 8088, AFAIK.

Oh puh-lease, I fucking worked for IBM in 1988. Despite your bullshit, everyone knew at the time that it was shortsighted not to design Windows with multi-user functionality in mind, and at the time, compatibility from version to version was broken anyway. They had a special version of Windows that worked on the 80286, and another one that worked on the 80386, and people basically used it as a shell to run a handful of apps. The big leap was Windows 95, and the decisions underlying the basic design of Windows 95 were made at a time when people did NOT need magic clairvoyance to see the future in order to know that shitty multiuser functionality would someday be an issue.

I think MS at least partly recognised the problems in that area, as they tried to start a new codebase with NT 3.1 and do proper multi-user stuff from day one. Unfortunately they fucked up right at the start by including needless hangovers from DOS, even though compatibility between DOS and NT was always going to be horrible.

Darth Wong wrote:They had a special version of Windows that worked on the 80286, and another one that worked on the 80386, and people basically used it as a shell to run a handful of apps.

Pardon me if I get this wrong, but didn't both versions of 2.x (Windows/286 and Windows/386) support Windows 1.x programs? Also didn't 3.0 (and latter 3.1 et all) support programs written for both version of 2.x?

Good luck with that. Most corporations like getting paid for their products and owning the rights to their products. I'm not going to say this is impossible, as I'm aware that a there different business models surrounding OSS, but its going to be an uphill battle to change this.

By License, I meant, a license required to be legally allowed to use a computer system, like a driver's license, not software licenses. Even open source software is licensed.

Office 2003 works great with Vista, and I have it my machine right now. Of course, Linux still does nothing for games or proprietary CE devices.

Well I've never tried to use an iPod with Linux, but PC-BSD is known for being able to read/write to iPods out of the box. Considering that's BSD, which is notorious for bad hardware detection, I'd assume its the same with Linux.

Linux webserves tend to be well administrated.

I dare you to try and proove that.

Also, regarding Arrows hardware choices, I'm suprised the above configuration didn't work. Perhaps it was the fact that it was so new-a major focus with Linux has always been on accomodating older hardware.

RThurmont wrote:By License, I meant, a license required to be legally allowed to use a computer system, like a driver's license, not software licenses. Even open source software is licensed.

Sorry, I got my arguments confused. Look, something on par with a driver's license is over kill, but some level of training is needed, and it should be required in school for graduation. I have a feeling some communities already have this, but my high school offered only typing courses and a basic programming course. It had nothing about basic computer maintenance.

Well I've never tried to use an iPod with Linux, but PC-BSD is known for being able to read/write to iPods out of the box. Considering that's BSD, which is notorious for bad hardware detection, I'd assume its the same with Linux.

But that doesn't cover a whole bunch of other devices.

Linux webserves tend to be well administrated.

I dare you to try and proove that.

I have only anecdotal evidence from friends and coworkers. Still waiting on proof from you about Joe Sixpack.

Sorry, I got my arguments confused. Look, something on par with a driver's license is over kill, but some level of training is needed, and it should be required in school for graduation. I have a feeling some communities already have this, but my high school offered only typing courses and a basic programming course. It had nothing about basic computer maintenance.

Well, I think kids should be given that kind of education in elementary or middle school. Teach them how to -use- a system in first grade, and by the time they hit sixth grade, make sure they also know how to -admin- one, to a simple degree.

RThurmont wrote:

Sorry, I got my arguments confused. Look, something on par with a driver's license is over kill, but some level of training is needed, and it should be required in school for graduation. I have a feeling some communities already have this, but my high school offered only typing courses and a basic programming course. It had nothing about basic computer maintenance.

Well, I think kids should be given that kind of education in elementary or middle school. Teach them how to -use- a system in first grade, and by the time they hit sixth grade, make sure they also know how to -admin- one, to a simple degree.

You'd face a very bitter battle in the curriculum departments over just where in the year the kids should take the time to get into these sorts of lessons. You'd also face a budgetary battle as I can assure you the majority of teachers would be simply unable to effectively teach such a course, and as such would require a new full-time position at each building for a computer teacher. Again, another bitter battle between 'do we hire regular classroom teachers and lower class sizes' (I'm not saying I'm subscribing to that philosophy, but it would come up) or do we add these tech teachers at the middle schools and high schools where the kids are already all consolidated?

Don't get me wrong. I work at a school district which I think has an obscenely weak technology curriculum: there isn't even a tech teacher at the high school, let alone any programming or sysadmin courses. (What's ridiculous is that there used to be classes like that, which then got axed.) I definitely think there ought to be good instruction on how to use a computer. But I think that a strong elementary-school tech program would be politically infeasible, especially because there would simply be too many people who would argue that it's not important for the student to know how his computer works so much as it is important that he's able to type out papers on it.

I don't think knowing how to program or administrate a system should be compulsory*, but these days I think students really ought to have a much better idea of how to use a computer (think more hints and tips for home use than run a network). The number of times I had to give pupils at my old school pointers (and that was one with a good IT department etc) was scary. I think once you've done the basic things in regedit etc though most people won't have a need to go much further.

*If you're into that sort of thing then by all means go to classes for it.

MariusRoi wrote:

Darth Wong wrote:They had a special version of Windows that worked on the 80286, and another one that worked on the 80386, and people basically used it as a shell to run a handful of apps.

Pardon me if I get this wrong, but didn't both versions of 2.x (Windows/286 and Windows/386) support Windows 1.x programs? Also didn't 3.0 (and latter 3.1 et all) support programs written for both version of 2.x?

In theory yes, but anyone who actually used those versions of Windows at the time will know that:

A) You only started Windows in order to run one of the one or two Windows apps you might have on your machine.

B) Windows was so horribly unstable at the time that you would get out to DOS as soon as you were done using that particular app. It was literally more like a runtime graphics toolkit (like GTK) than an operating system at the time.

C) The installed software base was 99% DOS, not Windows. People used DOS spreadsheets, DOS word processors, DOS games, the whole nine yards.

D) When you upgraded to a new version of Windows, you generally upgraded your one or two Windows apps along with it, because new versions came out around the same rate.

There was a huge opportunity to make a clean break with DOS conventions when they went to Windows 95, and they blew it. Everyone knew it was a mistake at the time. You had people writing articles in PC Magazine saying so, but today I have to put up with apologists pretending that you needed a fucking crystal ball to see it at the time.

RThurmont wrote:The file system is a part of the operating system, actually, and a part that is mmore often than not, not easily interchangeable. For example, with Windows XP you get a choice between FAT and NTFS, and with Vista, you just get NTFS (there might be a workaround for this, but it would probably be a general bad idea).

File systems are completely independent of operating systems. That Microsoft marries Windows to NTFS is irrelevant. As long as the operating system has drivers for a particular file system, you can use it.

The standard filesystems with UNIX like operating systems, or at a minimum, with Linux, do not have a fragmentation problem, thus, this is actually an advantage on their side.

Since when? Fragmentation is just a fact of life. HFS defrags small files on-the-fly, but I'm not aware of anything similar on ext3 or UFS.

OS X can be slow as refrigerated honey on my Mac Mini (Core Duo, purchased in September, the lower end model).

The low-end Mac mini is a Core Solo. I've run OS X on a Core Duo mini with a gig or RAM, and it ran beautifully. Faster than my dual G5 in some areas, especially the UI.

Yes there is. A system with proprietary APIs is always less desirable than a system with open APIs, as if the publisher goes out of business, or the product ceases to be supported, you're left high and dry, without the possibility to port to another platform.

In such a scenario, you'd have to end up porting the vendor-specific APIs over yourself anyway.

Solaris remained highly interoperable with other UNIX-like operating systems during that time frame.

As is Mac OS X.

Even today, closed source UNIX like operating systems such as Unixware and AIX are able to run without difficulty a huge range of software, like the "AMP" stack.

As is Mac OS X.

UNIX like operating systems share a number of characteristics, including file formats, file system arrangement, and so forth, and most of them can trace their lineage directly back to UNIX, anyway.

The characteristics they share are defined in the POSIX standard, which anyone can implement in any way they choose.

The Mac OS X, to be fair, is compatible with a lot of general UNIX stuff, but rather than building on the open standards it uses at its foundation, it instead supplements them with layer after layer of closed, proprietary technology, which Apple can and does discontinue support for at its leisure (see OS 9 apps and OS X on the Intel Macs, meaning Mac users can no longer enjoy classic Myst, among other tragedies).

CoreFoundation is open source, as I said before. Cocoa and the drawing layer are closed. But OS X is still plenty compatible with XWindows and POSIX applications.

As for Classic, cry me a river. The engineering effort to port Classic to Intel would've been massive, and there was no gain, and Classic would've continued to be an albatross around Apple's neck in terms of support. Besides, Classic has absolutely nothing to do with Mac OS X's Unix underpinnings. It's an entirely different operating system.

Darth Wong wrote:There was a huge opportunity to make a clean break with DOS conventions when they went to Windows 95, and they blew it. Everyone knew it was a mistake at the time. You had people writing articles in PC Magazine saying so, but today I have to put up with apologists pretending that you needed a fucking crystal ball to see it at the time.

Well, that would've been a technically-sound decision, but what do you do with all those legacy Win16 and DOS applications? People still wanted to run their old applications with Windows 95 (I know my family did) and a clean break would mean all sorts of compromises. NT went the route of virtualizing them but the average home machine at the time wasn't quite powerful enough to run that with acceptable performance.

phongn wrote:

Darth Wong wrote:There was a huge opportunity to make a clean break with DOS conventions when they went to Windows 95, and they blew it. Everyone knew it was a mistake at the time. You had people writing articles in PC Magazine saying so, but today I have to put up with apologists pretending that you needed a fucking crystal ball to see it at the time.

Well, that would've been a technically-sound decision, but what do you do with all those legacy Win16 and DOS applications? People still wanted to run their old applications with Windows 95 (I know my family did) and a clean break would mean all sorts of compromises. NT went the route of virtualizing them but the average home machine at the time wasn't quite powerful enough to run that with acceptable performance.

Those old versions of Windows had MS-DOS mode, where compatibility was guaranteed by actually suspending Windows and running DOS. Compatibility was guaranteed that way. But the directory layout and read/write permissions in Windows could have easily been configured differently. The fact is that Microsoft actually encouraged early app writers to mix app files with OS files, user data with app files, etc. They did it themselves with many of their early apps.

Darth Wong wrote:Those old versions of Windows had MS-DOS mode, where compatibility was guaranteed by actually suspending Windows and running DOS. Compatibility was guaranteed that way. But the directory layout and read/write permissions in Windows could have easily been configured differently.

I suppose you could do something like creating a globally-accessible sandbox directory and "fool" DOS applications into thinking that's the only real drive. Presumably you'd do the same with Win16 applications. I don't know well that'd work with users, though, especially if they want to store their stuff in their own directory.

The fact is that Microsoft actually encouraged early app writers to mix app files with OS files, user data with app files, etc. They did it themselves with many of their early apps.

True enough. I think we all remember DLL Hell (ugh!)

Well I've never tried to use an iPod with Linux, but PC-BSD is known for being able to read/write to iPods out of the box. Considering that's BSD, which is notorious for bad hardware detection, I'd assume its the same with Linux.

Linux and iPods work just fine together, since they're read as standard mountable USB devices. All you need is an application that can read the database (gtkpod, Amarok) and you're set.

File systems are completely independent of operating systems.

The filesystem that an operating system actually runs on is a core aspect of that operating system. While it can be interchangeable, the fact remains, that the default filesystems that Windows runs on are prone to fragmentation (FAT moreso than NTFS), whereas the filesystems that UNIX like operating systems typically run on are not.

The low-end Mac mini is a Core Solo.

No its not, its a Core Duo, and has been since September (albeit with 512mb rather than 1 gig of RAM IIRC). Anyway, you argued that OS X would be slow if:

Yeah, if you're running it on 4 year-old hardware. All currently-shipping Macs run OS X beautifully, and the OS has only gotten faster on older hardware with each release.

However, my Mac Mini is an example of a currently shipping Mac that runs OS X in a manner somewhat akin (at times) to Windows 95 on a 386.

RThurmont wrote:

File systems are completely independent of operating systems.

The filesystem that an operating system actually runs on is a core aspect of that operating system. While it can be interchangeable, the fact remains, that the default filesystems that Windows runs on are prone to fragmentation (FAT moreso than NTFS), whereas the filesystems that UNIX like operating systems typically run on are not.

You do realize that Linux based operating systems use FAT32 and some more recently use NTFS, yes?

You do realize that Linux based operating systems use FAT32 and some more recently use NTFS, yes?

I think ext3 is the Linux standard now. It was the default for Ubuntu anyway.

General Zod wrote:

RThurmont wrote:

File systems are completely independent of operating systems.

The filesystem that an operating system actually runs on is a core aspect of that operating system. While it can be interchangeable, the fact remains, that the default filesystems that Windows runs on are prone to fragmentation (FAT moreso than NTFS), whereas the filesystems that UNIX like operating systems typically run on are not.

You do realize that Linux based operating systems use FAT32 and some more recently use NTFS, yes?

Are there Linux distros which actually allow you to use FAT32 for the operating system partition? That would be an incredibly stupid decision.

Bounty wrote:

You do realize that Linux based operating systems use FAT32 and some more recently use NTFS, yes?

I think ext3 is the Linux standard now. It was the default for Ubuntu anyway.

I suspect he is thinking of the fact that Linux allows you to use pretty much any filesystem you want, when you're formatting a new partition. But I've never heard of a Linux distro that allowed you to use FAT32 for the partition which holds the operating system files.

Darth Wong wrote:

General Zod wrote:

RThurmont wrote: The filesystem that an operating system actually runs on is a core aspect of that operating system. While it can be interchangeable, the fact remains, that the default filesystems that Windows runs on are prone to fragmentation (FAT moreso than NTFS), whereas the filesystems that UNIX like operating systems typically run on are not.

You do realize that Linux based operating systems use FAT32 and some more recently use NTFS, yes?

Are there Linux distros which actually allow you to use FAT32 for the operating system partition? That would be an incredibly stupid decision.

I was able to format a hard drive using FAT32 with an Ubuntu distro some time ago when I created partitions for the Swap file and the other directories. Though admittedly I didn't know too terribly much about Linux at the time when I did so and wound up nuking it a few days later for Windows 2k.

I was able to format a hard drive using FAT32 with an Ubuntu distro some time ago when I created partitions for the Swap file and the other directories. Though admittedly I didn't know too terribly much about Linux at the time when I did so and wound up nuking it a few days later for Windows 2k.

That does not equate to Linux distros normally using FAT as the default filesystem type.

RThurmont wrote:The filesystem that an operating system actually runs on is a core aspect of that operating system.

No, it isn't. Operating systems are responsible for certain things. File I/O is one of those things, but the operating system communicates with the file system. The file system is not a part of the operating system. So your statement that Linux is better than Windows because of fragmentation doesn't make any sense.

While it can be interchangeable, the fact remains, that the default filesystems that Windows runs on are prone to fragmentation (FAT moreso than NTFS), whereas the filesystems that UNIX like operating systems typically run on are not.

No, you just don't understand computers.

However, my Mac Mini is an example of a currently shipping Mac that runs OS X in a manner somewhat akin (at times) to Windows 95 on a 386.

At times? That's hardly a basis to say that the whole OS is "slow". As I said before, for average use, any shipping Mac will run OS X very well. You shouldn't buy a low-end machine and then be surprised when it gets slow at times.