Offtopic rants (was: Why is nVidia So Problematic?)

If the desktop OS cannot do a straight forward install of Nvidia without tinkering then screw that OS. nVidia is a major player and I demand it install and run without hacking.
You need to ask Nvidia to release open-source drivers so FreeBSD can put it in the source tree.
FreeBSD can hack around their blobs so much but its not perfect. This is essentially out of our hands.

If you like BSD and want to keep with it, then check out NetBSD, they have Nouveau, the open-source Nvidia driver in their tree and it works very well.
 
y goodness (Maturin) but you are a condescending ass. Leave the personal insults to yourself as they have no place here. I do not appreciate them.
And I do, or what?

Ass - sometimes.
condescending - for sure not. Never, ever.
I always try very hard to have understanding for even the most stupid BS and taking effort to explain it all the way even a five ear old could unerstand while being concerned to stay patient, polite, understanding and gentle.
Insult? Where did I wrote a single insult?
YOU insulted me and others SEVERAL times! Just as the most recent example (again):
condescending ass
You are the fanboy
Do you read anything like that in any of my posts?
No. Because there are no such things.
Do you read your own posts?
Lost contact to reality?
Are on drugs?
What insulted you?
The truth?
The fact that you keep on mourning some old Windows situation you cannot have back, neither anybody here can fix it for you?
Or is it the fact that you are neither really satisfied with any Linux you tried, but at the same time keep up complaining here about the issues you have with FreeBSD, while not willing to learn it, which would prove:
FreeBSD ain't not tinkering, as I also told you many times.
If that insults you, you do have the issues.
Your problem, buddy, is, you are deeply unsatisfied with the situation, and neither willing to accept you cannot change it, nor anybody else can do it for you, no matter how much you complain, nor willing to be flexible enough to adapt yourself to a new system and stay with it.
But that's neither ones fault except your owns.
I learned myself FreeBSD, and learned, when you're doing it right (Handbook), it's exactly what you're looking for: solid, no tinkering.
I told you that several times - you don't wanna hear it. That's not my fault.
You do not want to accept this. That's is not my fault.
Pointing out, that's contradicting and pointless makes me a
condescending ass
and a
and being accused of insults.
Take my last advice:
Seek help.
You true issues are neiter with any OS, nor some NVIDIA drivers, or anything like it.
It's something like a dissatisfaction you carry somewhere deep inside yourself.
But this here is neither the place, nor any other technical forum, for helping you with that.
Only you can solve your problems yourself, nobody else can do that for you.
All this forums here can provide, is helping you with you some certain technical issues.
But you don't ask for help, at least do not take any advice.
All you do is complain, whine, mourne and tell some "if it was otherwise" BS.
And if somebody points that out, especially to tell you, that also doesn't help your case neither, maybe to reconsider it all over completely new again, you start a brawl, insulting others while accuse them of insults they have not made etc.

Ignore list.
Permanent.
While you neither need, nor want my help, it's no loss to you neither.
 
you are deeply unsatisfied with the situation, and neither willing to accept you cannot change it,
IDK, seems like they accepted that they were't going to use FreeBSD to solve the problem 😆

I think it's reasonable to expect that nvidia hardware work without having to fiddle with it, particularly if Linux does, and be annoyed that it doesn't.
 
I think it's reasonable to expect that nvidia hardware work without having to fiddle with it, particularly if Linux does, and be annoyed that it doesn't.
It's been stated twice, and now a third time,
You need to ask Nvidia to release open-source drivers so FreeBSD can put it in the source tree.
FreeBSD can hack around their blobs so much but its not perfect. This is essentially out of our hands.
 
I’ve been tinkering in this business 53 years now and simply want to do work without constant tinkering. I’m tired of having to tweak things to make them work.
If you don't want to tweak things, don't run the latest release. Stay on 15.0 until it is retired. 15.0 is what I have on a pi3 (since 15.1 libc tickles a hardware bug).

However, you should consider donating to the freebsd project. With more money more things will get fixed....
 
If you don't want to tweak things, don't run the latest release. Stay on 15.0 until it is retired.
Pointless.
I (and others) already told him things like that several times.
All you get by him is, 'It just works with [some Linux distro].', 'I don't like tinkering' and, 'Windows were all I just need.' etc. Well, OK then.
But when you tell him, it's just well OK then, so simply stick to it then, you are accused of being an condescending ass insulting (by being insulted.)
Pointless.
And pointing out, that's pointless, just give it at least a shot, makes you being the bad guy being accused a "fanboy insulting others".
You just have to live with, that bgavin is just here to complain about FreeBSD does not come as a full automated fool-proof turn-key installing his favortite GUI DE of choice without the need of placing a single simple shell command, while also avoiding GhostBSD, which would provide exactly that, and mourning, that the great days of exemplary Windows 95 are over.
(Sorry, had some wine when I wrote this - I AM pissed. Trying to be patient, understanding, helping and being accused of doing an insult, while there was none [I read my post 5 times; can't find any.] can makes you pissed.)
 
Pointless.
I (and others) already told him things like that several times.
All you get by him is, 'It just works with [some Linux distro].', 'I don't like tinkering' and, 'Windows were all I just need.' etc. Well, OK then.
But when you tell him, it's just well OK then, so simply stick to it then, you are accused of being an condescending ass insulting (by being insulted.)
Pointless.
And pointing out, that's pointless, just give it at least a shot, makes you being the bad guy being accused a "fanboy insulting others".
You just have to live with, that bgavin is just here to complain about FreeBSD does not come as a full automated fool-proof turn-key installing his favortite GUI DE of choice without the need of placing a single simple shell command, while also avoiding GhostBSD, which would provide exactly that, and mourning, that the great days of exemplary Windows 95 are over.
(Sorry, had some wine when I wrote this - I AM pissed. Trying to be patient, understanding, helping and being accused of doing an insult, while there was none [I read my post 5 times; can't find any.] can makes you pissed.)
I'm tempted to request a turnkey system that:
  • Rebuilt everything from source on base upgrading, by clicking YES on common-to-DEs dialog box querying "Updates for base FreeBSD is found. Do you want to upgrade now?" and proceed unattended pulling of targetted srcv tree / rebuilds / installkernel if appropreate / installworld / forcible rebot when YES is clicked.
  • Same for ports.
It should be a complete turnkey computer after first installation is finished.
windows-update often takes hours on long standing, oldest supported hardwares. So taking hours / days SHALL not matter here. Isn't it?

Maybe the largest hurdle to do it would be per-field merges of /etc instead of current line-based merges by etcupdate(8) and/or now-removed (on 15.0) mergemaster(8).

Note that I've once proposed (actually, requested, as I don't think I can implement it myself) to make kmod pkgs to be a kind of hooks to build kmods locally but contains everything else just as normal pkgs. The answer was, as you can see, providing current FreeBSD-{ports-}kmods repo instead.
 
Maybe the largest hurdle to do it would be per-field merges of /etc instead of current line-based merges by etcupdate(8) and/or now-removed (on 15.0) mergemaster(8).
That's what I also see, as the biggest problem.
All the rest can be fully automated with scripts.
My idea so far was - if someone wants to create a FreeBSD based turn-key distribution, do such things by yourself (the maintainer) once manually as a template, provide them by a central servers (doesn't need much space, a few kB at most), and the according machines can get it from there, e.g. by a script run by root's cron as part of the update procedure.

But, when you think it through - and I did, since also within my friends there are several ones not happy with Windows anymore, but not knowing shit about computers (Windows only users), neither willing to learn the smallest bit, nor caring at all (e.g. which DE, as long as it's not too strange) - apart from doing this as some excerzise/brain teaser/hobby, I see no point in starting another fork while there already is a pretty useful solution as far as I can judge already providing the most jobs for a good turn-key OS based on FreeSD:
GhostBSD
In my eyes it would make way more sense to join that project, and put such effort in there, instead of splitting working power, and having smaller projects, eventually lacking of people to consequently maintain it.
Don't forget: It needs to be reliable. Not some Linux student's project,: halfhearted started, and never got even remotely to some useful state before interest was lost.

Anyway the largest hurdle I see is independent from any BSD, Linux or what else.
It's every Window leaver must face the fact to learn and use other Software.
In many cases it's simple: Use firefox instead of explorer, thunderbid (or something else) instead of outlook, LibreOffice instead of... you get the picture. Don't underestimate, that's already very much for the common Win user, spoilt with eyecandy, never seen such software.
And there is at least a must to learn at least some small bits about computering - one must at least learn how to install/deinstall software packages.
And as you can see almost every month here, pkg install pdfviewerXYZ is already too much for many.

But what concerns me above all at most is, there is lots of Software in Windows Universe, which cannot so easily be replaced. I'm not taliking games. Cut games! Focus (first) on the users not playing.
I'm talking all this Microsoft 356 Meeting collaborating and video calling stuff.
 
More than video stuff, I'm thinking of what I'd term professional software because it's used by what are commonly called professionals, that is, doctors, dentists, lawyers, accountants, and the like. For example, dentists have all sorts of programs that only run on Windows, Same with doctors, and I'd guess lawyers, etc., as well. Where I am, Zoom is adequate for video calling,and that can work in a browser, but the various complex medical and other software is going to be a huge barrier to many.
 
More than video stuff, I'm thinking of what I'd term professional software because it's used by what are commonly called professionals, that is, doctors, dentists, lawyers, accountants, and the like. For example, dentists have all sorts of programs that only run on Windows, Same with doctors, and I'd guess lawyers, etc., as well. Where I am, Zoom is adequate for video calling,and that can work in a browser, but the various complex medical and other software is going to be a huge barrier to many.
Before Windows dominated PC markets, what application softwares are provided for specific OS was the largest reason for users to choose OS.
The next may be "is my boards in my PC supported by this OS and runs stabully and fastest among others?".
 
but the various complex medical and other software is going to be a huge barrier to many.
Most medical offices and hospitals have a contractor that delivers and maintain everything. It's completely unimportant which OS or hardware brands the stuff runs on, as long as it works. And if not, an technician is called to make it run asap again.

But one has to start somewhere, and private uses without gaming are already mire than enough machines.

Before Windows dominated PC markets, what application softwares are provided for specific OS was the largest reason for users to choose OS.
That's still the case today. A doctor in a hospital just needs to use his hospital software.
Companies need their accountant software, which OS it's running on doesn't matter.
Today it's just simple: Since everybody runs Windows, every Software is written for Windows, so everybody runs Windows.

The next may be "is my boards in my PC supported by this OS and runs stabully and fastest among others?".
The next will be everything is kind in a browser; hardware doesn't matter much, so doesn't the OS, as long as it's capable to provide a GUI which connects to some online servers ("cloud compzuting" or what's it called) - MS is almost completely there.
But here we - the open source OS - become more interesting. At least in the private sectors. Because MS is overdoing things, which becomes a growing nuisance to more and more users, also in public areas. And it's also political. France is developing their complete own system for their government offices Linux based.
 
The next will be everything is kind in a browser; hardware doesn't matter much, so doesn't the OS, as long as it's capable to provide a GUI which connects to some online servers ("cloud compzuting" or what's it called) - MS is almost completely there.
It's not just MS: you have applications like QuickBooks and Adobe that really really push you do their online version. They also muck with licenses: lets say you've been using QB for a business and you are winding the buisiness down. You still have a need to access at least read-only for some period after closing (taxes); well last time I asked Intuit says Nope, not even read only if you don't have a license.
That is huge problem.
 
Ok, well, step one in breaking the hegemony is being able to install an OS that can responsively draw pretty pictures on the screen without having to know the model number of the GPU in the process...
 
You need to ask Nvidia to release open-source drivers so FreeBSD can put it in the source tree.
FreeBSD can hack around their blobs so much but its not perfect. This is essentially out of our hands.
Give up and perfect the shim until we're Tier 1 for them 🤷‍♀️ Wireless was important enough to do this.
 
Give up and perfect the shim until we're Tier 1 for them 🤷‍♀️ Wireless was important enough to do this.
The wireless LinuxKPI shim works because we have the source code to the Linux iwlwifi driver. Shims don't unfortunately work with blobs. Otherwise I would agree this is the way forward.

In light of this, I'd recommend give up and go full in with nouveau until we're tier 1 for Nvidia.

The blob from Nvidia is lacking cuda, lacks support for older hardware compared to nouveau and requires a shed load of patches and compromises.

That said, we will never be tier 1 for Nvidia. We might as well start planning around this now. Once they drop the blob, we will be dead in the water.
 
Most medical offices and hospitals have a contractor that delivers and maintain everything. It's completely unimportant which OS or hardware brands the stuff runs on, as long as it works.
That's actually true for most non-computer professionals. The laptop/desktop machine has become a pure appliance, used only to "run software", more on that below. The choice of OS is a decision internal to the IT infrastructure, which doesn't affect users much.

The next will be everything is kind in a browser; hardware doesn't matter much, so doesn't the OS, as long as it's capable to provide a GUI which ...
... which provides a standard browser. Which is why Chromebooks are such good computing platforms: No overhead for running apps locally. Running everything "in the cloud" (with the user interface being through a browser) makes administering IT so much easier and more efficient. To begin with, you don't have to worry about data stored on end user devices. Nor about compatibility, as long as they can run a common browser.

I used to work in what today is probably the largest software company in the world (the one with the most software engineers). Lots of my software engineer colleagues used Chromebooks. They would a browser-based ssh when they needed to log in to a CLI, they would use browser-based graphical tools for code editing (I preferred emacs in a CLI window, but not everyone does), they would use graphical tools for CI/CD and code review workflows, for database access and authentication, graphical dashboards for data access and monitoring, and so on. I had a Mac (partly because the keyboard is more ergonomic for my damaged hand, partly because the screen is bigger for my old eyes), but a lot of the others had Chromebooks. In this environment, it doesn't matter at all what the OS on the device is ... it doesn't matter that the Chromebook runs some version of Linux or Android, while my MacBook runs MacOS, as long as Chrome and a text window work.

In my opinion, for end user computing, the era of the operating system is over. It's now an unimportant technical detail.
 
The wireless LinuxKPI shim works because we have the source code to the Linux iwlwifi driver. Shims don't unfortunately work with blobs. Otherwise I would agree this is the way forward.

In light of this, I'd recommend give up and go full in with nouveau until we're tier 1 for Nvidia.

The blob from Nvidia is lacking cuda, lacks support for older hardware compared to nouveau and requires a shed load of patches and compromises.

That said, we will never be tier 1 for Nvidia. We might as well start planning around this now. Once they drop the blob, we will be dead in the water.
I agree with this, i.e. also start supporting nouveau and nvk, it would mean a more consistent support starting with g80 cards (point at which modesetting works fine). NVK is also progressing nicely.
The performance is not on par with the official drivers but at least post maxwell it's no longer painful to use.

I've tried starting this but time, skills and hardware are not on my side.
 
In my opinion, for end user computing, the era of the operating system is over. It's now an unimportant technical detail.

As I said within other correlations, one must not see such things by its technical points, only.

By the pure theoretical technical user's point of view, this was very desirable, makes perfect sense.
Nobody needs to care about having data saved securely. Neither need for learning how or doing backups. Not even the need for the most rudimentary computer knowledge, like saving, copying, renaming files, installing/deinstalling software etc. No need to be considered about hardware failures. No need to care for any hardware anyway (enough memory, storage space, CPU/GPU power; no different storage mediums are needed anymore, etc. etc.) No need to think of and learn anything about computers at all, just the usage of the UI of the application software, only.
Short: Just use the computer what's it actually all for, and forget about all the rest.
Plus you could safe a lot of hardware. This way you don't need to buy the most recent top state of technology machine for each user every two..three years, which would save a lot of resources, and money. It could be done easily, and since it's very desirable it's sold that way, but is not really done, because then the hardware manufacturers had losses. But this leads too far away from my point I'm trying to make here.

Btw I love this irony: All those old Unix machines until the 1980s have been a centralized computer with terminals attached to it. After all offices in all companies have been stuffed since the 1990s with complete fully equipped computers just to connect to the accounting software on the central server, we finally coming back to that concept of terminals. Embedded into the background noise of "greybeards telling war stories from the stone ages, while they shall contemporize to the modern world." 😂

All this is based on and leading to even more centralization, while everybody with some natural scientific based education, every engineer who's even remotely worth his title knows, distributed system are more solid, more healthy, more safe against disturbances, because they are more flexible, independent, adaptable. They do not all fail completely when the center fails.
Everything is always very nice as long as the shit works. But everybody not lost contact to reality knows: No real existing system always runs 100% flawless. Every engineer knows, one does not plan for its flawless work. One must plan for exceptions, problems, breakdowns, disturbances, changes, adaptions, maintenence, accidents... to make the shit work as flawless as possible.

Already the practical technical issues can cause severe issues: The bandwidth of internet ain't infinite. If internet connections fails, nothing goes at all anymore. For example I am in an area where constantly new buildings are constructed, and old ones are teared down. Everytime a house is added to the glass fiber net, the local net's capacity needs to be extended, or an excavator hits a line, leaving apart a power line, we have no internet connection for a couple of hours or even days - which of course you don't know in advance. Sudden Strike: Without a warning all the shit just stops working. Such happen every couple of months. If you never experienced something like it, you may say, 'Every couple of months for a few hours. Oh please, come on!' Well, nothing goes at all. And sometimes even for days. And everybody ever did any payed work knows, there are time schedules must be met, most of the times "tomorrow", sometimes even deadlines. And the common working day of the common employee ain't intended to have a couple of hours free time.
There is no need to remind here, what a failure of the central server means.

Which brings us to two of the core concepts in systems analysis:
The pros and cons of centrality vs. distribution.
And the possibility of planning, the predictability decreases when the complexness increases.
Centralisation doesn't neccasserily mean a decrease of complexness. It only means a transformation and relocation of the according modules a system consists of. Complexness does not primarily come from centralized or distributed. It mostly comes from the tasks a system needs to do. The inceased complexness a distributed system brings is not decreased just by centralising. The parts from which the complexness of a distributed systems comes are not all elimated by centralisation. Instead you get new parts, adding new, other, unkown complexness. Sometimes the complexness of a system is even increased by centralisation.
At the same time centralisation feels pretty like a system becomes less complex that way. This is a trap. Because of all the modules, their interaction and collaboration are summarized into fewer pieces of hardware, so we see fewer things. As a reflex our brains tell us, 'Fewer things we see equals less complexness.' That's simply not true. Especially not, if most of the modules a system consists of are software. Software running in the background without showing an UI, so invisible and unkown to the common computer user.

But as I said, don't look at the technical points, only.
There are also the social, business, political and other points need to be respected.

The lack of computer knowledge creates dependency.
Centralisation creates more control, surveillance, power.
And that's why it's tried again and over again. Not to make the world a better place, but to milk the cattle even more.
And that's exactly why we need to preserve OS like BSD or Linux, to offer, preserve and guarantee alternatives, when this centralisation finally fails. And it will. It always does.
It's pure nature: When things are overdone they break.
And that's exactly the problem with capitalism: joint-stock companies MUST always increase their wins constantly. Otherwise their indexes rushing down the toilet. But when you reached the limits of growth - we're all living in a limited system, no matter what scifi BS some YT crap makes some dream of - the methods of how to grow even more and fighting the rising numbers of knock-backs, which are immanent to the situation trying to overdo the limits, finally become absurd.
And that's exactly what were seeing right now. Not only with MS, who are already milking their cattle that hard, the cows start complaining.
 
And from the point of view of criminal crackers (don't call them hackers!), centralized system may be most desirable.
They need to know about only 1 system (computer and softwares in all layers).
Even if per-user data stored there are encrypted with separate per-user keys, they have a plenty of time to crack once they can clone a bunch of data.

On the other hand, diversed computing with a quite wide variety of OS'es, middlewares, applications, ... make criminal's work harder.
If all users of target organizations has different computers (CPU, firmware, OS, ...) except for authentication protocol and decrypted data formats, they are forced to determine the environment of actual target user, look for vulnerabilities for the specific hardwares, firmwares, OS and softwares running on it (even software for authentication can be different, as what's needed to fix is the protocol, and in many cases vulnerabilities in protocols are fewer than ones in implementations), then, attack when the computer is up and running with Internet connections.

Yes, the trade-offs between maintainability/manageability and safety, if each users are aware enough for security of their computers.

But for centralization, isn't mainframe the mandatory minimum requirement as hardware and OS? (Not z/Linux but z/OS, for example.)
 
Back
Top