For what reason FreeBSD in some directions so slowly develops?

Status
Not open for further replies.
Recently I had a feeling that opensource-products become more and more habitual in an everyday life.
Behind examples it isn't necessary to go far:
  • Huge ecosystem of Android of decisions are products for entertainments, communication, continuous use and they make fantastic success.
  • Leading game studios are guided by development of games as well for Linux of clients
  • Producers of the equipment in the majority promote support of the products in Linux
And it everything certainly very good tendencies, but as very old user of FreeBSD guards me the fact of that our system very strongly lags behind in some directions of development the advanced opensource of decisions (Linux and Android based on it).

I chose for myself FreeBSD that developers of this system don't go on compromises between quality of a system and introduction of innovations, but at the moment the situation looks disturbing:
  • When wireless decisions, are so actively used - FreeBSD is very weak in support of the wireless equipment (no .n support, weak many device support).
  • When specifications on the equipment long ago are open and support is perfectly realized in Linux, in FreeBSD with it everything is bad, or in any way. (videocards, TV tuners)
  • When conducting opensource projects introduce new ideas, FreeBSD is not ready to that these products worked in it (new Mesa, X.org, v4l2, GEM,KMS,TTM support for new videocards except Intel)
Separately I am frightened that often opensource products let out only-linux, i.e. even in the Posix-world there is any binding to one system.

Reasons of it in my opinion are as follows:
  • Insufficient financing (but why not to declare in End-of-Year Fundraising Campaign separately financing any obviously lagging behind and necessary for active use of system of technologies? I suspect grants development very good way to find and stimulate developers)
  • Insufficient quantity of developers (Can be it is worth carrying out more aggressive marketing or system promoting? ) . For example: All of us (users of FreeBSD) know, what it is very good system for network tasks but why still not to find in sale a router or L3-switch on the basis of FreeBSD except Juniper products, but at least most it is almost impossible to have opportunity перепрошить the device on FreeBSD?

While already there are too much questions and theses in one message therefore I will finish.
At the end I want to tell:
I very much want that FreeBSD which system can use for the solution of any tasks. Also it is desirable that it could do not only system programmers who at big desire can simply realize that is necessary for them in system.
 
And why the community of FreeBSD doesn't arrive in the same way if it yields positive results?
 
It seems to me that the answer: It not so - it isn't constructive.
I have nothing against FreeBSD, I use only it in the work but if to look objectively,
  • Built-in systems - Linux
  • Virtualization (Jail is good but it not virtualization, and virtualBox it is frivolous) - Linux
  • Support of the wireless equipment - at FreeBSD practically everything is very bad (very few drivers, aren't present.n support)
  • New technologies - also in FreeBSD appear with a big delay after Linux
I understand and completely agree that treat development of FreeBSD much more responsibly than in Linux, and also I see that the result of development turns out much stabler, logical, clear and predictable than in case of Linux of systems.

And so I don't understand one why at all pluses of FreeBSD in its development participate much less people and the companies and also significantly less resources than go to Linux.
 
The vast majority of computer users not adequately informed? In which case a few flowcharts could be useful (Setup firewall, printing, email, desktop...). Here you have a lack of time from present FreeBSD users AFAIK.
 
G_Nerc said:
It seems to me that the answer: It not so - it isn't constructive.
I have nothing against FreeBSD, I use only it in the work but if to look objectively,
  • Built-in systems - Linux
  • Virtualization (Jail is good but it not virtualization, and virtualBox it is frivolous) - Linux
  • Support of the wireless equipment - at FreeBSD practically everything is very bad (very few drivers, aren't present.n support)
  • New technologies - also in FreeBSD appear with a big delay after Linux
I understand and completely agree that treat development of FreeBSD much more responsibly than in Linux, and also I see that the result of development turns out much stabler, logical, clear and predictable than in case of Linux of systems.

And so I don't understand one why at all pluses of FreeBSD in its development participate much less people and the companies and also significantly less resources than go to Linux.

They tried to do it and failed with two previous releases. Here will give you some insight; btw I'm glad they acknowledge the problems they had( I lived through them) :

http://www.freebsd.org/doc/en/articles/version-guide/past-schedules.html

http://www.tr.freebsd.org/doc/en_US.ISO8859-1/articles/5-roadmap/index.html

I can't answer the last question. In fact no one can. It's almost bizarre how GNU/Linux got so popular. I hear the same thing with references to LISP. Where the language may have been usurped by the marketing of the new fast business language COBOL.

It is a strange world we live in.

Maybe this silly video may give some insight:
https://www.youtube.com/watch?v=bzkRVzciAZg
 
G_Nerc said:
  • Support of the wireless equipment - at FreeBSD practically everything is very bad (very few drivers, aren't present.n support)
  • New technologies - also in FreeBSD appear with a big delay after Linux
Support for such things, in most cases, is up to the hardware vendor that created them and not FreeBSD or Linux. While someone somewhere may take the challenge to write their own driver, it's frequently hit and miss and not as good as the original. If wireless equipment doesn't work in FreeBSD, ask the manufacturer of that device why they didn't supply a driver.
 
One way FreeBSD may be falling behind is the install phase. During a typical install and configuration, there should be no room for error. Many freebsd books and web guides are already obsoleted in part... which is why I suggested a flowchart, it could be dynamically updated and simultaneously encourage those migrating from linux and windows, that they would not have to compile a mini-howto before working straight from a (expensive to produce in time and energy) flowchart. For instance, just within the past week I first had a chance to finish a working GPT disk install, stumbling through a combination of other cli and ncurses tools (non-GPT installs) year after year. No way I could deduce the steps I took plainly from the man page. A flowchart, if large and verbose enough, could provide more examples and show a larger picture and provide a larger preplan than a manpage might, at least in my opinion.
 
jb_fvwm2 said:
One way FreeBSD may be falling behind is the install phase.

Agreed. OpenBSD and NetBSD have an easier path when installing. It seems more "closed" and less prone to errors. I like the OpenBSD one especially.

Regarding to the thread author's points. I do not understand what he says about virtualization, and as for new technologies ZFS sounds quite advanced to me.
 
A few points:

1. FreeBSD, for mostly historical reasons, has a lot less people working on it. They are mostly volunteers, so what gets changed tends to be what they need. Companies sometimes fund development for specific features they want, but it's not common. The point of this is that if you want something, push for it. Make it your own.

2. 802.11n support is in 10.0, at least for some Atheros and Intel cards. I don't know if it has been or will be MFCed to 9-STABLE.

3. An analogy: FreeBSD is to Linux as PostgreSQL is to MySQL. Different goals, different attitudes.
 
Many companies and schools- colleges, universities- are either unaware of FreeBSD and the other BSD's as well; or, people do not always want to learn something new. Costly people hours will be used as a reason. Of course, I don't know how many peoples it takes to make an hour.
 
Installed PC-BSD 9.1 last night. I was impressed. Installer was very straightforward, auto detected my X setup, my sound setup, etc. To the extent that a splash video played before first login.

I'm sure there's rough edges (haven't played with it much yet), but as far as end user desktops go, it's a step in the right direction.
 
drhowarddrfine said:
If wireless equipment doesn't work in FreeBSD, ask the manufacturer of that device why they didn't supply a driver.

Really? We all know why... The market for FreeBSD is so miniscule, why would the manufacturer spend time and energy writing drivers for it?

Besides, it's not just wireless equipment... My HD6950 is next to useless in FreeBSD. Sure, I can use the Xorg vesa driver, but why would I? If I were to ask AMD about FreeBSD support, their response would be that they don't write drivers for FreeBSD because the user base is so small. On the other hand, they'd point out that they do have their own employees working on open source drivers, pay contractors to work on said drivers, and release specifications. They'd continue that there's no one stopping developers from using the code and specifications to make the open source drivers work on FreeBSD.

Clearly, however, that's not a priority of the FreeBSD project. Which is fine. Using FreeBSD on my desktop or laptop is no longer an option for me, then.

Adam
 
The fact that your new trendy wifi card doesnt work is not a sign that FreeBSD is progressing slowly. That same card probably works fine on Windows 2000 with the correct driver, does that mean that OS is more modern?

If your hardware doesn't work on FreeBSD, replace it with working components. This is afterall what Apple does. Mac OS X supports much less hardware and yet I never hear anyone complaining that their fingerprint reader etc... doesn't work.

Whilst I do agree that FreeBSD is developed at a slower pace in some places, it is also developing at a much higher pace in others for two reasons:

1) Due to the license stuff, we could cannibalize a lot of Solaris tech.
2) Linux reinvents the wheel (systemd) every few years, so regresses back to the start. It gives a false illusion of progress.

Other than hipster crap like flash, FreeBSD is easily ready for the desktop of most power users in the same way that Linux is.
 
I have one more question, why FreeBSD developers not developes drivers based on Linux drivers? I know that GPL and BSD license are radically different, but why FreeBSD developers can't see Linux drivers and write drivers for FreeBSD (not grab from Linux)?
 
kpedersen said:
If your hardware doesn't work on FreeBSD, replace it with working components.

So what are my options for a GPU manufacturer that is friendly to open source operating systems? Please note, this is not the same thing as releasing closed source drivers for FreeBSD.

Adam
 
G_Nerc said:
write drivers for FreeBSD (not grab from Linux)?

It is often easier to port an existing driver than write a new one from scratch. Afterall, many of the magic numbers that hardware needs is the same across platforms.

However, as is the nature with Open-source, Linux has taken its fair share of BSD driver code. I believe that many Linux wifi drivers originated from the OpenBSD project.

Not to mention that much of the network stack in Windows was originally borrowed from BSD ;)

adamk said:
So what are my options for a GPU manufacturer that is friendly to open source operating systems? Please note, this is not the same thing as releasing closed source drivers for FreeBSD.
Pretty much the same options as any other open-source operating system... Pretty dismal unless you are using an older intel card haha.
For any serious 3D development on Linux, you are going to be using the nvidia binary blob. Likewise on FreeBSD.

Where Linux is apparently further is with Nouveau and Gallium projects but frankly (As someone who develops using OpenGL) they are not yet usable because when debugging my own software (especially shaders), I don't know if the problem is due to my code or the fact that the underlying driver is broken.

I recently had an issue with Nouveau where I could only pass 2 uniforms into the fragment shader, the others were completely ignored (although they were fine in the vertex shader). It took me sodding hours to work out the issue and then install the nvidia binary blob ;)

Admittedly Linux is also further ahead with KMS and newer intel drivers... But again, if the newer hardware doesnt work, dont use it. I am using an Intel GMA 965 and it works fine on FreeBSD (just not very quickly).
 
G_Nerc said:
I have one more question, why FreeBSD developers not developes drivers based on Linux drivers? I know that GPL and BSD license are radically different, but why FreeBSD developers can't see Linux drivers and write drivers for FreeBSD (not grab from Linux)?

Don't kid yourself, Linux grabs code from other operating systems too. Wrong place to nitpick/bash BSD.
 
kpedersen said:
Pretty much the same options as any other open-source operating system... Pretty dismal unless you are using an older intel card haha.

Really, because my HD6950 does everything I need it to under Linux with the open source drivers. Even Haiku, of all things, has a developer working on modesetting for radeon HD GPUs, with a number of people reporting success.

For any serious 3D development on Linux, you are going to be using the nvidia binary blob. Likewise on FreeBSD.

On FreeBSD I'd settle for basic modesetting...

Where Linux is apparently further is with Nouveau and Gallium projects but frankly (As someone who develops using OpenGL) they are not yet usable because when debugging my own software (especially shaders), I don't know if the problem is due to my code or the fact that the underlying driver is broken.

As a user, I don't care :) I want to be able to plug in a very common video card, get two monitors working, and be able to watch at least low definition videos.

BTW, the developers (at least the radeon developers) are very responsive to questions from application developers about problems, and could probably easily tell you if the problem is with your code or the driver.

But again, if the newer hardware doesnt work, dont use it. I am using an Intel GMA 965 and it works fine on FreeBSD (just not very quickly).

So, again, what are my options for a video card from an open source friendly GPU manufacturer? nVidia is not an option since they aren't open source friendly, and Intel is not an option since I'm not going to replace my entire motherboard because I'd prefer to use FreeBSD. I shouldn't have to replace my motherboard just to use FreeBSD.

Like it or not, the very fact that we need to have this discussion shows that FreeBSD is, indeed, being developed slowly in some areas compared to other similar operating systems. Even OpenBSD had support for newer Intel GPUs long before FreeBSD.
 
I see what your saying and I would also like to use 100% open-source drivers and in another 10 years, I almost certainly will be.

However, rather than use Linux, I am quite happy to chuck out stuff that doesn't work and replace it with stuff that does work on FreeBSD. In other words, I choose the operating system and then the hardware to match, rather than the other way round.

My solution is to use binary blob nvidia driver for 3D intensive stuff / development. And then use the old Intel GMA which comes with most Thinkpads for day to day stuff. This should be quite satisfactory for 99% of users.

Does your HD6950 not get quite warm when running under the open-source driver?
I had a HD4xxx which ran warm under the Gallium drivers but then ran cool on the fglrx binaries but had a really slow 2D redraw. This is the sort of crap, I don't want to spend time with on Linux.

adamk said:
Even OpenBSD had support for newer Intel GPUs long before FreeBSD.

Driver support is a never ending battle and really doesn't represent the development speed of an OS. In my opinion, it isn't the job of an OS to supply the drivers. Open-source operating systems currently have to purely because hardware manufacturers are ignorant.

Also, since the aftermath of Gnome 3, I have lost a lot of hope in the open-source philosophy. For example, even though Gnome 2 was open source, it is impossible for a single user to grab the source, compile it up and use it in place of Gnome 3.
It is the same with the Nouvea or Nvidia drivers. If the developers pack up and go home, Nouveau on Linux will probably disappear, regardless of if it is open-source or not in exactly the same way as the Nvidia binary drivers. The only benefit of open-source drivers (for me as an end user) is better integration to the OS.
 
adamk said:
The market for FreeBSD is so miniscule, why would the manufacturer spend time and energy writing drivers for it?
I really don't get this. I don't understand what's so difficult about porting a driver over to FreeBSD when one exists on Linux. I'm not saying there is no effort involved but come on. Maybe a high end 3D graphics card that needs something special only Linux has...OK...but a wireless card?

As a EE I used to design hardware and write the drivers. The drivers ran in a real-time OS we bought but we could switch OSes and rewrite the drivers with little effort. That was then and this is now but it still baffles me.

I've said this before. If I were younger, I'd start a hardware company that creates boards for FreeBSD and Linux exclusively. I'd bet there'd be a ton of money in that.
 
drhowarddrfine said:
I really don't get this. I don't understand what's so difficult about porting a driver over to FreeBSD when one exists on Linux. I'm not saying there is no effort involved but come on. Maybe a high end 3D graphics card that needs something special only Linux has...OK...but a wireless card?

I never said it was difficult, just that the market is so small it's not worth it.
 
kpedersen said:
However, rather than use Linux, I am quite happy to chuck out stuff that doesn't work and replace it with stuff that does work on FreeBSD. In other words, I choose the operating system and then the hardware to match, rather than the other way round.

If there were an option for me, as there was a few years ago, that would be fine.

My solution is to use binary blob nvidia driver for 3D intensive stuff / development. And then use the old Intel GMA which comes with most Thinkpads for day to day stuff. This should be quite satisfactory for 99% of users.

99% of users just buy common hardware and.expect it to work.

Does your HD6950 not get quite warm when running under the open-source driver?
I had a HD4xxx which ran warm under the Gallium drivers but then ran cool omn the fglrx binaries but had a really slow 2D redraw. This is the sort of crap, I don't want to spend time with on Linux.

I never saw this problem.

Driver support is a never ending battle and really doesn't represent the development speed of an OS. In my opinion, it isn't the job of an OS to supply the drivers. Open-source operating systems currently have to purely because hardware manufacturers are ignorant.

They are profit driven, not ignorant. It's ignorant to use an operating system with such a small user
base and expect hardware manufacturers (especially ones already releasing source code and specs) to write drivers for that
OS.

Adam
 
Well, those comments were certainly uplifting and invigorating.

I believe we shall call this the thread of apathy.
 
Status
Not open for further replies.
Back
Top