OSDev.org

Posted: Mon Dec 23, 2019 3:57 am

by PeyTy

GreenteaOS goal is to run unmodified Windows .exe apps. Not here yet :D I made lots of .exe loader prototypes, now finishing ACPI parser and will finally implement public version of .exe loader.

Wiki article - https://wiki.osdev.org/User:Peyty/GreenteaOS
GitHub - https://github.com/GreenteaOS (I would be so happy if you would star some repos :roll: )

Code is weirdly styled due to me not liking C++ so much :mrgreen: but actually to simplify future back-port to Hexa, my another complicated project - a systems programmering language.

You may join discussion and/or development at Telegram https://t.me/greenteaos and Discord https://discord.gg/UGZq8GB

Ask me anything! :!:

Screenshot is from old dirty prototype. I hadn't time to work on clean VFS implementation to post on GitHub.

Image

Posted: Mon Dec 23, 2019 5:09 am

by bzt

PeyTy wrote:GreenteaOS goal is to run unmodified Windows .exe apps.

An admirable goal, but let me remind you that the wine project ("windows emulator", which has exactly the same goal), has several developers and after a decade of coding it is still not perfect nor finished by any means. This is mostly because Windows is not documented at this level, only the upper-level API is public.

Also there's another OS with exactly the same goal, ReactOS, not surprisingly unfinished too and keeps crashing all the time. Admittedly they also aim for device driver compatibility too, which is a lot more problematic and more complex task. They have made very impressive progress recently, and they are in close collaboration with the wine project, but still it is not enough to create a stable system (unfortunately, I hope they will succeed one day).

Both of these projects have considerable advantage over your GreenteaOS, most notably decades of experience with reverse-engineering low-level WinAPI and lot more developers. So what makes you think you alone can do it better?

(Don't get me wrong, you have a very noble goal, but knowing the difficulty it's just a bit far-fetched I think.)

Cheers,
bzt

Posted: Mon Dec 23, 2019 5:41 am

by PeyTy

bzt wrote:...ReactOS, not surprisingly unfinished too and keeps crashing all the time...

...and the reason why GreenteaOS were created. I posted a link to wiki where you may read exactly about that.

bzt wrote:...decades of experience with reverse-engineering...

That sounds like "common wisdom", but reverse-engineering is not that hard and Wine/ROS are open-source so experience may be shared.

bzt wrote:...knowing the difficulty...

I'm trying to be extremely creative in development approaches. I do see that other projects... well.. I mean, the only existing project on the planet like this? ReactOS struggle to make progress with their "driver compatibility" approach. Greentea tries to do it different way: make working solid userspace and only then consider about hardware stuff. Also, Greentea aims at only x86-64 (2009+ CPUs) so it also greatly simplifies kernel design. On the other side, ReactOS uses C which is pretty much useless for implementing NT, NT is object design written in C++. I cannot be 100% sure what is better that's why I'm just doing it other way and see how it goes in practice.

bzt wrote:considerable advantage over your GreenteaOS

Well Wine is not OS and Linux not even remotely considered by a lot A LOT like hundreds of millions of users.
ReactOS has only conceptual advantage - it is not even beta.
GreenteaOS existence is 100% okay.

I know, it every-time ends with GreenteaOS vs ReactOS debate, but really, those are intentionally very different projects with similar goals.

bzt wrote:you alone

Open-source is not alone :) I already have some devs who help me to design the kernel.

Thanks for you interest! Sorry if my English may sound aggressive at times. I'm working on GreenteaOS 4 years already and it's ok for me to keep trying until notable results achieved.

Posted: Mon Dec 23, 2019 9:30 am

by bzt

PeyTy wrote:I cannot be 100% sure what is better that's why I'm just doing it other way and see how it goes in practice.

Now sir, that is a perfectly valid reason to do this! I wish you good luck with your project, let's hope you'll be more successful than ReactOS!

PeyTy wrote:Open-source is not alone :)

A link to the source maybe? You have mentioned github, but without any repo links.

Cheers,
bzt

Posted: Mon Dec 23, 2019 11:48 am

by thomtl

@bzt Link is on the wiki page, https://github.com/GreenteaOS

Posted: Mon Dec 23, 2019 12:55 pm

by PeyTy

@bzt thank you for the kind words!

@thomtl added link to the post, thanks

Posted: Fri Dec 27, 2019 7:06 pm

by eekee

Speaking as a recent convert to Windows, I think this looks interesting! I'm all in favour of cutting out tough unnecessary parts: drivers and 32-bit support in this case. I look forward to seeing how you get on!

With LeafDOS announced on the 19th and GreenteaOS on the 23rd, I'm wondering what's next. FruitOS the Mac clone? LovebirdOS the Amiga clone? That's getting away from plants. And LeafDOS isn't a clone anyway. Mine isn't either. Maybe I should call it PepperminTeaOS. :lol: This all reminds me of the cream soda computer I read about once.

Posted: Sun Dec 29, 2019 2:58 am

by codepilot

I read that you want to add NVMe support. I am working on putting very simple how to on the osdev wiki for NVMe with some working code.

I am writting it in a few different languages, C++ being first

Posted: Sun Dec 29, 2019 4:01 am

by PeyTy

@eekee thanks for the GitHub stars and support!

codepilot wrote:I read that you want to add NVMe support. I am working on putting very simple how to on the osdev wiki for NVMe with some working code.

I am writting it in a few different languages, C++ being first

Yes, would love to see your article. NVMe is now supported by VirtualBox and I have hardware to test it. Protocol is simpler and less demanding than SATA, natural choice to implement it first.

Posted: Sun Dec 29, 2019 11:00 am

by codepilot

PeyTy wrote:Protocol is simpler and less demanding than SATA, natural choice to implement it first.

If i never see sata again, i would smile :)

Posted: Sun Dec 29, 2019 12:06 pm

by mmdmine

the most big problem of windows is that it's a big code spagetti. it lacks a unique design. you may get same error for same reason in different styles (3.1, xp, 8) at once. what makes windows slow is its compalibility with older softwares and hardwares. i like what you are doing, and i would use it when it got stable. good luck.

Posted: Sun Dec 29, 2019 3:11 pm

by zaval

mmdmine wrote: the most big problem of windows is that it's a big code spagetti. it lacks a unique design.

did you see that code? or it's just GNU/beliefs are telling from you. if you wanna look at the huge, messy, unstructured, poorly thought out pile of code, that claims to be an OS, then it's name starts from li and ends with nux. and no, it's not GNU/linux, stallman may keep eating his toe cheese, it's still just linux. should linux have at least 10% of the uniqueness of Windows, it would be a thing. not just ugly, non-innovative, uninspiring unix clone. I love, when people parrot such BS you did about things they have no idea about. "problem" of windows, "lack" of uniqueness, "code" spaghetti, "slow"... oh, really, I smell FSF "expert" on Windows "problems". you forgot yet about "viruses" and "proprietary". libre is endangered, world needs moar open sauce, only fsf crackpots and their religion know what people need in their computer. and this is frigging freedom! under gpl3+. jeez.

you may get same error for same reason in different styles (3.1, xp, 8) at once. what makes windows slow is its compalibility with older softwares and hardwares. i like what you are doing, and i would use it when it got stable. good luck.

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. not fearing to say that, got sick of these open sauce fairy tales. it's not true, I convinced myself a million times, how opposite to reality all such claims you did are. you may dislike MS as much as your religion requires, but they have done the most efficient OS and it deservingly won the desktop market. and it's concrete fact. and I like it because it is the best in this - efficiency. (I like its GUI most as well and as well, gnu/libre/free/open/schmopen stuff sucks turbulently in this field). that's why it pisses me off so much, when I see texts like this. lying.

but well, move on, to the topic. this greentea thing, interesting the author is silent about its relationship with reactos. and it's strange, because I am sure, reactos people have a few words to say about this greentea os.* in short, it's a rip off, pretending to be something independent. did you change your mind and started your thingy from scratch or still copy-and-pasting reactos code, pretending it's not derived from it?
finally, NT is object based, but it's written in C. I heard Cutler hated unix but liked C and I am not sure he felt the same to C++.

* - here, even sitting on this lame android tablet, with sadistic "keyboard", being nearly useless thanks to the "brilliant" OS from the Corporation of Goodness, that is as "free" as crappy, it almost always hangs, running on a dual core processor and with 1 GB of RAM, sigh, so, still, I went and made a little search on the ros forum. interesting reading, PeyTy. one, two, three.

Posted: Sun Dec 29, 2019 4:04 pm

by Schol-R-LEA

zaval wrote:In reality, Windows is the most efficiently implemented general purpose OS I have been faced with.

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.

As for Windows, many of the knocks against it are due to historical and pragmatic issues rather than design or implementation flaws; many of the same have held for systems such as Mac System (prior to OS X), despite that starting out as a genuinely efficient system in several ways (albeit an overburdened and unstable one) with the remarkable work done by Hertzfeld on the ROM Toolbox and the frankly insane ways that they tuned and tweaked things such as sound generation and video timing on the 128K Macintosh. Linux has plenty such problems, too, but really, hype to the contrary, its primary niche is a very different one anyway (with the overwhelming majority of installed systems being either mobile devices or servers, mainly for economic rather than technical reasons - though portability across architectures, and fine-grained tuneability for HPC requirements, are a big part of it).

Many of the problems which have plagued Windows - quite a few of which no longer do - came from trying to blend four radically different OS designs (CP/M by way of MS-DOS with the original Windows as an add-on, bits of UNIX mixed in from MS-DOS 2.0 onwards, OS/2 as developed when MS and Big Blue were still working together, and the glorified re-write of VMS which became Windows NT) into a coherent whole as they moved from the 9x kernel to the XP kernel, all while trying to maintain compatibility with programs dating back upwards of 20 years written with the assumption that they could access the hardware directly any way they chose.

That, and the utterly blase attitude which the consumer market in general holds towards such things as stability and security, again for reasons having nothing to do with technology and everything to do with cost and convenience. In this, Microsoft was just giving their customers what they wanted, regardless of whether it was what they needed or not. They are hardly alone in this; IMAO, it is little short of criminal that registered ECC didn't become the standard for system memory from the day it was introduced, but raw speed and low cost are apparently much better selling points for Intel than actually staying stable enough to finish what you were doing without crashing.

Further, Microsoft's perennial excuse that third party drivers are often at fault is not entirely without merit, though it was a lot more true before Microsoft introduced a certification process for hardware vendors who wanted Microsoft to give their customers the time of day on support issues. For a long time, Linux developers simply didn't support - or feel a need to support, since 'you can always write your own driver' - the breadth and variety of hardware which Windows did, which made it significantly easier to make the kernel stable for the hardware it would run on. Similarly, the kernel team only needed to provide a basic level of support for a family of platforms - x86, x86-64, ARM, MIPS - and leave the rest of the work adapting it to, say, a new SBC or smartphone onto the hardware vendors and their consumer supporters.

Finally, the sheer size of the installed base for Windows has made it a bigger target for those looking to attack, undermine, or compromise the system for their own gain. If one looks at the places where Linux derivatives are dominant - such as Android devices - one soon sees that many of the same problems which plague Windows affect them just as badly.

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.

Posted: Sun Dec 29, 2019 4:42 pm

by zaval

schol-r-lea, sorry I need a code monkey from google near me, to use their tablet, they said it will dismiss PCs, I tried to use it now, it costed to me tons of nerves and a half an hour to just barely paste three links into the text, and check it's done right, it's pure sadists, who made this, not even lamers. apparently I need one of those dudes stating this is the future to type instead of me. would be fun to make all google to sit and do their work on such tablets. and as well, again I don't want to argue, for me what I said is undisputable and needs no farther discussion. my bad, that I failed to not let my anger out. I know, I am right there but I needed not to argue with that BS. and also, even though this greentea is fake, still my arguments wouldn't be on topic here.

as a bonus, it drained battery while I was fighting to post here. on a phpbb forum, with plain text! holy guacamole.

calling NT VMS rewrite is like calling Itanium an Alpha rewrite, just because there the same people worked. but both are not. farther evolvement yes, because it's the architect's ideas that he puts into his new things are evolving. it was not a "steal", as MS hating fanatics say. if you read about Files-11 and NTFS, it momentarily would be obvious that calling one rewrite of another is well overstatement. elephant is not a rewrite of mommoth. MS-DOS was brought up to satisfy demand (IBM's) for quick OS solution, MS didn't consider it a serious design for years. WinAPI has been transferred from Windows to NT due to its obvious growing popularity, and it landed there pretty well as we see, so nothing awkward. IBM demanding to not do that was awkward, however understandable, they didn't want MS own the API. well, we know who won. and yet, I like to repeat this, Gordon Letwin in his 95 letter stated OS/2 was started at MS as a "serious" OS for PC, after they ditched Xenix. then due to the split they went an NT way throwing away their own initial plan (OS/2). and that NT thing is still doing very well. So what we have now there is a pretty consistent platform, that won both internally and externally the competition and that comes with both good kernel design and API. nothing grottesque, byzantine or whatever.

Posted: Sun Dec 29, 2019 5:37 pm

by portasynthinca3

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.