OSDev.org

portasynthinca3 wrote:Excuse me for being off-topic here, but I think you both have just trashed the original topic (a new OS with very cool goals) and replaced it with yet another "Windows vs Linux" dispute.

The irony of this is that this was precisely what I was trying to stave off, by pointing out that most of the 'problems' of Windows are either consequences of its history, or shared by all other current OSes, or (in most instances) both. I should know better than to reply to someone whose only real interest is in arguing. I am every bit as fed up with both bzt and Zaval as I am with Trident/Manbaby/QuantumRubber and ~, and quite honestly I wonder why I am still here at all.

zaval wrote:your fantasies make windows slow. in your head. In reality, Windows is the most efficiently implemented general purpose OS I have been faced with. all this open sauce crap is mediocre and lame sh1t in comparion

You made the classic mistake to mix a kernel with the operating system.

Let's get his straight:
1. Windows (as an OS) IS slow, like a pregnant snail, all users know that, it is a fact.
2. the Windows kernel is one of the most efficiently implemented kernel (the kernel, not the OS)

The difference comes from mainly these factors:
1. after M$ failed to create a real operating system (with Win 95/98/and specially ME), they hired the guys who wrote VMS
2. designed by real experts, the NT kernel therefore is really good
3. however M$ failed to share the documentation with the developers all around the world on how to use the NT kernel or the surrounding DLL jungle efficiently, so most user-space programs are really bad and ineffective
4. as an overall result, the end-user experience is terrible, and people are only buying it because they don't know better.

Calling Open Source mediocre and lame just proves that you're not smart enough to use them. No worries, not everybody is ready to be woken up from the Matrix

This reminds me of the old classic joke: what would it be like if OSes were flight companies?

Windows NT: beautiful stuardesses, nice looking airports with lots of friendly signs to guide you, all using Cleartype typefaces, however all seems to lead to the most expensive lobby. Everybody assures you that there'll be no problem, but sometimes planes crash. You often travel the 90% of the full distance in 80% of the time, then you spent the remaining 10% distance with 30% of the time for some uncertain reasons, and nobody knows why.

Linux: no stuardesses at all, just freely downloadable how-tos, plain simple airports with just arrows and bitmap typefaces to guide you, but to the right direction in the shortest path. When you get on the plain, the pilot gives you a screwdriver and tells you to set your chair as you like. If there's a problem, the plain doesn't even get off, no air disasters. The flight is smooth and predictable. You arrive in time without any problems. But later, when you try to explain your amazing journey to your friends, they just keep asking "...and you did what with the screwdriver...?"

So I really would like to see a free and Open Source operating system that looks like the Windows NT airport but is as predictable and transparent as the Linux flight. I really think they should persue this project, maybe something great come out of it!

@Schol-R-LEA: yes, my post on this topic was only about arguing.

Cheers,
bzt

I was going to join in, but then I remembered everything I know about Windows internals was told to me by a VMS expert.

I do have a little experience, from which I'll say: if you give Windows plenty of memory, it's fine. I have no idea why it wants so much. I had some other efficiency complaints, but they're all consigned to history now.

I'm not commenting positively or negatively on other systems in this thread, except of cource GreenTeaOS.

Interesting links though. I do rather hope this is not just a case of Dunning-Kruger effect, although even those can be educational for the participants.

@zaval keep calm. i use windows and i like it. nothing is perfect. both linux and windows have their own problems. one of problems of windows is its design.
i didnt see its code, but i can feel it when i use it. windows 10 is based on 8 and it's based on 7 and ... till NT that designed to be compatible with dos and windows 9x. all of them are different from each other but still you can see windows 3.1-style dialogs in windows 10 while 10 has its own new style.
ms prefers compalibility on user exprience.
windows 10 has both control panel and its new settings app, but one is enough. although settings is not complete and still you need control panel for some configurations.
or try to run a 16-bit exe (on 64-bit windows 10) once from explorer once from a 32-bit program (like 32-bit 7-zip fm) and once from cmd to get same error in 3 styles.
windows has many signs of bad design, that could be solved if they manage codebase correctly.
your right, many open sources softwares are not as user friendly and stable as windows and its softwares. but they are faster. as an example i can tell you that my archlinux installation boots in 6 seconds but windows takes at least one minute. sometimes i prefer speed and sometimes i prefer windows' feel. so i use both.

zaval wrote:fsf religion ... this greentea thing, interesting the author is silent about its relationship with reactos

ReactOS is same kind of religion. I was heavily trolled in every imaginable way. You may just ignore whatever ros fans are saying about GreenteaOS. I decided to end this insanity and start own thing from scratch and absolutely never ever worry about that project. Nowadays Greentea shares no code with reactos.

portasynthinca3 wrote:Excuse me for being off-topic here, but I think you both have just trashed the original topic (a new OS with very cool goals) and replaced it with yet another "Windows vs Linux" dispute.

Here is for hoping someday someone adds a third contender into the ring

codepilot wrote:Here is for hoping someday someone adds a third contender into the ring

Well, there is FreeBSD. And OpenBSD, NetBSD. And Mac OS X in so many iterations.

I have enough experience to support GreenteaOS and understand ReactOS proponents feelings at the same time. Too many times, I was the idiot shouting down something new, often due to faulty or overly-extended reasoning. I'm sort-of trying to make up for that now. My opinions were largely based on perceived self-interest, but I even managed to misunderstand my own needs.

On another note entirely, I wasn't going to talk about other OSs, but bzt's plane seat illustration is too much fun to resist! To fit myself into the allegory, Linux plane seats have too many necessary screws, making my injured neck hurt too much as I contort myself under the chair to do them up. BSDs have fewer screws to actually fit the chair securely, but the adjustments to make it comfortable are just as bad as Linux. Plan 9 has the fewest screws of all, and some adjustments are far easier than in any Unix, but other adjustments require you bring your own engineering workshop!

Mac OS X I saw at its best, 10.4 & 10.5, where I'd say it combined much of the best of Windows and Unix. Beautiful stewardesses[1] and an optional screwdriver to fit Unix seats. The plane has some unnecessary slowdowns unless it is large (RAM). They seem to intentionally make older planes very slow when newer planes are available. 10.3 was actually bad; it was like 60% of flights didn't reach their destination.

[1]: Surely they must have a discriminatory hiring policy!

Schol-R-LEA wrote: I'm sorely tempted to say that this is because you're too young to have known any actually efficient operating systems, but then, the same applies to me, too. Note that the ones I have in mind as being efficient designs - such as ITS, Tenex, and TOPS-20 - are long since gone. So much for efficiency being a major factor in success.
Unix never really was one of them; it was merely so lightweight and minimalist in its early days that it could squeeze onto inexpensive systems such as the PDP-11, at the cost of not being able to do anything much with it except teach programming classes.
Besides, and efficiency and clarity of design are generally at odds, rather than complementary. There are exceptions to this rule, but far too few.
...
That having been said, I am in no mood to defend any of the existing operating systems, as they are all fundamentally and catastrophically bad. Unfortunately, they also are catastrophes which we are stuck with for the foreseeable future, and no one here is likely to change that fact.

While efficiency is an issue, there is also functionality as an operating system that is lacking. Gang scheduling seems to have been forgotten, alongside memory scheduling, the latter of whose loss leads to thrashing and inferior global replacement schemes for hobbyist end users. I think kernel-level clustering support didn't make it from VMS into Windows NT, though I'm pretty sure genuine asynchronous IO did. I think full DSHM/DIPC/DFS with process migration hasn't been seen since Miloječić' patches for Mach, and those weren't even included with Mach 3.0, never mind the performance problems Mach had that L4 "resolved" while throwing NORMA overboard. While various kernels pay lip service to superpages, Navarro et al's full scheme hasn't really gotten into many if any kernels, nor have I seen any papers regarding conceptual improvements on the idea. I suspect a great deal of good could be done by simply resuscitating various features or design goals from the history of operating systems in combination, pursuing quality of implementation, and maintaining the result over time.

While my impression is that this forum is geared more toward what undergraduate operating system classes used to be than research kernels for the next HotOS or USENIX conferences, I'd certainly love to see some of the latter.

nyc wrote: Gang scheduling
memory scheduling
kernel-level clustering
DSHM/DIPC/DFS
NORMA
superpages
Navarro et al's scheme

Any links? I hadn't recognised those terms (at least in that context) and guess many people didn't...

PeyTy wrote: Any links? I hadn't recognised those terms (at least in that context) and guess many people didn't...

• Gang scheduling: https://en.wikipedia.org/wiki/Gang_scheduling
• Memory scheduling: https://en.wikipedia.org/wiki/Working_set -- the paper explaining the scheduling aspect of it, http://web.stanford.edu/~ouster/cgi-bin ... duling.pdf is also referenced in the coscheduling and gang scheduling Wikipedia articles
• Kernel-level clustering & DSHM/DIPS/DFS: https://en.wikipedia.org/wiki/Single_system_image explains at least some of it
  • DSHM = Distributed SHared Memory
  • DIPC = Distributed IPC
  • DFS = Distributed File System
• NORMA: https://en.wikipedia.org/wiki/NoRMA
• Superpages / Navarro et al's scheme: https://en.wikipedia.org/wiki/Page_(com ... page_sizes or see the original paper at https://people.mpi-sws.org/~druschel/pu ... rpages.pdf

There are plenty of other research topics in operating systems e.g. filesystems can go into advanced data structures and algorithms and scheduling and networking often go into the weeds of queueing theory.

PeyTy wrote: I decided to end this insanity and start own thing from scratch and absolutely never ever worry about that project. Nowadays Greentea shares no code with reactos.

that's another story then, good luck with yiur affair.

and to bzt, I just saw your reaction to my post here. frankly it's not worth even mentioning, but I'll make one note, because well, I am stupid idiot, that even knowing crackpots should be ingored will be bickering in response. certainly not the best side of me. *shrugs*. so. everything you wrote there is as lie, incorrect and BS as your M$ writing or that moronic "cheers" are. yet once you proved yourself being a BS generator. you parroted a well known fairy tail of GNU/morons - all the MS are lamers, there was just a bunch of real professionals they hired from DEC, who did a good work, but MS screwed it all and >90% of people use Windows because they don't know better... yeah, right, very smart. and original! go tell them about "better". dude, thinking this primitive brain farting is the truth calling me not smart enough to see "beauty" of linux. LOL! you sound as a 14 linux lover pumped with unconditional love to GNU/Stallman and hatred to MS. totally brainwashed, shallow and fanatical. your histerics about that patch in qemu among the moaning about removing your priceless parricipation after the ban are all from that category. you are an infantile nuthead. the only usefulness of you is when you throw links at that Xv6 thing. I could not worry less of what your opinion on Windows or anything on this planet is. I use it and linux and can compare by myself. without lunatics like you. see, a dude, seriously thinking Reiser was chased by FBI/CIA/aliens, thrown for killing his wife in jail, whom he hasn't killed, instead it's a ruSSian mob in collaboration with FBI/CIA/aliens committed and all this because he was a mega talented linux developer can not be taken seriously. Brendan was surely right banning you, as you are inclined to bring to the forum a lot of insanity.

to schol-whatever. I reacted here because also was fed up with that kind of statements, I reacted to. and I also get frustrated/disappointed/distracted by such things and don't want to come up here, so why are your "fed up" is better, than mine? and also I can't get how you can be fed up with me, when I never have been bickering/arguing with you?even didn't touch you. I just am not interested in things you are. those overacademical balderdash or (pre)historical things or that LISP - things absolutely not interesting to me. maybe you ingore me and be better?

nyc wrote:...

Thank you for your time. Investigated some links. To be honest, they don't look like a breakthrows in osdev. Every point has it's disadwantages. BTW superpages were implemented in a user space memory allocator (this one if I remember correctly - https://www.reddit.com/r/cpp/comments/a ... nt_for_cc/ )

zaval wrote:good luck with yiur affair

Thanks. Please, no personal attacks here

PeyTy wrote:

zaval wrote:good luck with your affair

Thanks. Please, no personal attacks here

of course, I also would be glad to read in threads like this more of (planned or implemented) features of the announced OSs, concepts they employ, methods/algorithms/approaches used etc, but announcers seem to be not inclined to be that talkative on that matter, so the threads end up being a 1-3 posts worth with "this is my OS. yay! good luck. thanks" content.
I don't really understand why people are so concerned with "personal" attacks, especially on the internet, for me they are not more severe, than attacks on the truth. for me when some linux wanker shows up and out of nothing busts "windows is slow as a pregnant snail and this is a fact" is a much much rude attack and insulting, than if the same linux wanker would call me a "moron". because I don't care about personal attacks from unknown people, whereas that shitting windows by linux wankers is just ... I am tired of it, it's raping the truth by the sect of psychopaths. it got too much. even wikipedia is a shithole with this respect. I don't understand these fanatics with their delusional "freedom" as well as their insane hatred. i don't understand how their opensauce is a world saving thing, I don't understand why I should hate a corporation, making software and the OS, that is a million times better, than their godawful, vomitting inducing "alternative". this movement, driven by the crazy is a toxic sh1t, contaminating people's minds, they turned into zombies that appear everywhere and demand everything "opensauce". it's depressing, especially because I see how mediocre stuff come out as a result of this "revolution". again, while I was typing this, my opensauce android ate up more than a half of battery charge. for f&cking typing in a textbox! typing, that is near to torturing. and then, you see an idiot bleating about "slow" OS from "M$"! this is a personal attack! I am done anyway, no more reaction to bzt and similar types. actually, lately there is too much bzt over here, so I'd better go anywhere else.

zaval wrote: of course, I also would be glad to read in threads like this more of (planned or implemented) features of the announced OSs, concepts they employ, methods/algorithms/approaches used etc, but announcers seem to be not inclined to be that talkative on that matter, so the threads end up being a 1-3 posts worth with "this is my OS. yay! good luck. thanks" content.

AFAIK your ANT "project" still has zero lines of code? Don't reproject your problems on other people.

I do talk about osdev A LOT but not on this specific forum. Final decisions result in a code and blog posts. Talking takes lots of time and not always productive. I also take many personal notes and read books.

And yes, "good luck. thanks" is a very nice addition to a life of developer...

zaval wrote: I don't really understand why people are so concerned with "personal" attacks, especially on the internet, for me

For you? Maybe. But you may ruin a day of another person. We are humans.

zaval wrote:...

You said nothing new. Windows has both pros and cons, and mostly a personal preference. Despite me doing Windows-like OS, I don't reject that many people do love Linux, macOS or whatever. And I will not agree that Windows is groundbreakingly better than Linux. I don't even implement "NT", after all, only Win32.

I would enjoy seeing your expertise in Win32\NT . How far did you go investigating ANT design?