Will systemd make FreeBSD more popular?

Status
Not open for further replies.
retrogamer said:
It would still stink to get left behind, but then what would happen to PC-BSD?
Click to expand...

PC-BSD wouldn't go anywhere if FreeBSD itself stopped offering half-baked (just my opinion) X11 based desktop environments that are barely usable for newcomers. PC-BSD could very well just take over the maintainance of the X11 and desktop environment related ports leaving FreeBSD in charge of the core technologies that are not X11 or DE related.
 
j7j3 said:
Funnily enough as a desktop/server user the only real gripe I have had with FreeBSD since I migrated from Linux is driver support for my other laptops horrible Optimus chipset
Click to expand...
Have you tried FreeBSD 10.1-stable on your Optimus laptop? I have recently installed it onto my Asus U35JC laptop (Optimus, check) and to my surprise it works quite well (FreeBSD 7, 8 and 9 never did).
 
http://homepage.ntlworld.com/jonathan.deboynepollard/FGA/launchd-on-bsd.html
The discomforting truth is that we aren't going to get launchd for doing service and system management for the very same reasons that we aren't going to get systemd for doing service and system management. systemd is full of Linuxisms. launchd is full of Machisms. It's simply not a BSD program. It's a Mach program.
Click to expand...

I don't think the BSDs need launchd or systemd, because diversity is good and one thing is not good for everyone. But where there is a will, there is a way. The engineers at Joyent are proof of this when they ported KVM to SmartOS; then Joyent resurrected Sun's one-time Branded Zones to release LX branded zones to run Linux applications directly on bare metal; and Joyent has ported Docker to SmartOS bare metal.

now that Open Indiana is pretty well dead
Click to expand...
Yes and no, OpenIndiana itself hasn't had a release since 151a9 in January 2014, however, the hipster branch of OI is alive, active and publishes application changes regularly along with iso releases.
 
Launchd adoption has already started a while ago. Integration/testing with TrueOS is already in the works so it's a matter of time before it gets feature completed. Now whether the FreeBSD project will accept and pull in changes is a whole another thing on its' own.

Exciting times indeed.
 
Beastie7 said:
Launchd adoption has already started a while ago. Integration/testing with TrueOS is already in the works so it's a matter of time before it gets feature completed. Now whether the FreeBSD project will accept and pull in changes is a whole another thing on its' own.

Exciting times indeed.
Click to expand...
IMHO, something designed from scratch would be a much better idea than porting over Launchd to FreeBSD if at some point that is indeed the direction the FreeBSD developers decide to go. FreeBSD != TrueOS, similar but still some what different project goals.
 
tingo said:
Have you tried FreeBSD 10.1-stable on your Optimus laptop? I have recently installed it onto my Asus U35JC laptop (Optimus, check) and to my surprise it works quite well (FreeBSD 7, 8 and 9 never did).
Click to expand...

I would if I could switch off the Intel card in the BIOS but alas I can't. I even got in contact with MSI to ask one of their BIOS engineers for a custom BIOS to be able to turn it off but the laptop is one of those awkward beasts that is not able to bypass the Intel chipset.
 
j7j3 said:
I would if I could switch off the Intel card in the BIOS but alas I can't. I even got in contact with MSI to ask one of their BIOS engineers for a custom BIOS to be able to turn it off but the laptop is one of those awkward beasts that is not able to bypass the Intel chipset.
Click to expand...
My point was that you might not need the "BIOS switch", if the driver works and does it for you. (My case is opposite; I turn off the nVidia card via acpi_call and a script, and just use the Intel chip to get longer battery time).
 
tingo said:
My point was that you might not need the "BIOS switch", if the driver works and does it for you. (My case is opposite; I turn off the nVidia card via acpi_call and a script, and just use the Intel chip to get longer battery time).
Click to expand...

Thanks for the tip I'll look into that.
 
gpatrick said:
Yes and no, OpenIndiana itself hasn't had a release since 151a9 in January 2014, however, the hipster branch of OI is alive, active and publishes application changes regularly along with iso releases.
Click to expand...
If you don't mind me asking, have you tried living with it recently, and if so, how stable was it? I don't have any intention of jumping ship, but I might play around with it on an older box that I used to run OpenSolaris on.
 
OpenIndiana has always been solidly stable, just there isn't any updates for security, etc., although one could do a nightly or regular build on their own to get the illumos changes. The last time I installed OI was last weekend because I intended to build it to a newer release, but then after building packages and building illumos I stopped.

As for hipster I've never installed it, but since it uses OI 151a9 I'd assume it to be just as stable, however, since they are regularly updating applications they mention it may break sometimes. But that is an application, OI would still be stable.
 
retrogamer said:
It would still stink to get left behind, but then what would happen to PC-BSD? I'm sure a community derivative would still spring up to maintain a desktop version of some kind, FreeBSD is the only good desktop option if you want ZFS now that Open Indiana is pretty well dead. It still makes the most sense of any BSD or Linux as a multimedia oriented desktop OS for a power user, I think, because if you want:

1. OSS that is still well supported
2. ZFS
3. Non-obfuscated init system
4. Decent graphics drivers
Click to expand...

I've come to really appreciate two operating systems for meeting your four "power user" requirements. One of course is {Free,PC}BSD. The other is Funtoo Linux. I multi-boot between FreeBSD and Funtoo Linux on ZFS with my home directories on a ZFS mirror. Both support ZFS very well and Funtoo uses a very clean init system, OpenRC. Funtoo has committed to work well for both servers and desktop and to continue using OpenRC as its default and preferred init system. I've been finding myself doing most of my work (octave/matlab, DrRacket, smlnj, haskell, vim/emacs/latex etc. for coursework and small projects) under Funtoo where I have less problems installing ports than I sometimes do under FreeBSD.

Funtoo Linux ZFS is slightly behind FreeBSD ZFS so create any shared ZFS pool in Funtoo first and do not upgrade it from FreeBSD. Funtoo Linux does seem to be a pocket of sanity. :)
 
The solution is to rewrite and not patch or to correct the code - a partial rewite - of the problems in the Linux/GNU system. Debian has been forked into Devuan by a group not happy with the decision to go with systemd. I understand if systemd was used as a layer between a chroot or virtualized environment; but, there is no reason to add unneeded complexity in a system. On smaller devices, this is a waste of valuable memory.
 
Well, Red Hat's decision to go all-in for SystemD with its code bloat, its being a possible SPF due to being a large program run as PID 1, its diversion from Unix philosophy, and its multiple tentacles into Linux distributions has our level III admins at my business seriously discussing how FreeBSD can be used to replace CentOS on our servers going forward. Most of our internet facing services are standard open source program stacks such as Apache, BIND, NTP, etcetera. None of which need Linux with SystemD to work properly. We are working up a business case to present to our department director.
 
There's several companies using FreeBSD that has (via videos, whitepapers) done comparative analysis, described their experience, and explained why they chose FreeBSD. Netflix, Whatsapp, Limelight Networks, and recently Groupon to name a few. I would use that and build a report to present to your superiors. You can even tweak your presentation to explain how FreeBSDs offerings are better for your environment.

If they're concerned about support, they can call iXsystems.

That's how I would "sell" FreeBSD at least.
 
UnixRocks said:
We are working up a business case to present to our department director.
Click to expand...
Good luck to you.

It always makes me sad when you need to work in this way. If your management only speaks 'powerpoint', things usually go downwards from there.
 
Thanks guys. TBH, we are also discussing just rolling out new systems with FreeBSD installed and not mentioning it to management at all. We administrators are the only ones that will be touching those systems anyway. One of the team pointed out that if we already have core infrastructure successfully running FreeBSD as a fait accompli before anyone notices, it will be pretty easy to "sell" at that point. Our director is primarily concerned about vendor support for specific enterprise apps that "require" we use either Red Hat or CentOS. For now we use CentOS for those. We plan to just keep those systems as-is, upgrade them as needed to newer CentOS and live with the SystemD pain on those few boxes / VMs. But we do not need Linux for the majority of the servers we run.

Sad anecdote: We are an ISP. My division handles the internet customer facing systems, BIND, NTP, Webmail / POP / IMAP, Speedtest, and so on. A different division handles the "corporate properties" and internal corporate networking. Some "bright" beanie wearer in that other division hired some outside contractors who placed significant portions of our web property on Microsoft's Azure cloud rather than using our own stuff. My team just recently found out about that when some of the ISP web properties we manage stopped working correctly when trying to pull content from the main site, which was having problems because of some issue with Microsoft's "cloud". That was a major "WTF?!" when we found out. Basically typical corporate SNAFU. ;)
 
Another Debian systemd refugee here. Systemd is not the only reason I switched but it's the big one. The proverbial straw. I debated switching to another Linux distribution that doesn't have systemd but at the rate that it's assimilating everything I believe those other distros will have not choice but to switch sooner or later.

So last night I took the plunge and installed FreeBSD on one of my two Linux boxes. The BIOS and SSD firmware updates meant I had to re-install anyways so it was the perfect time. The other Linux box needs some new hardware so again it will be the perfect time to replace it with FreeBSD when it's rebuilt.

So far I've manually did the root on ZFS procedure as I wanted 4K alignment. Today it was compiling a custom kernel. So far I'm very pleased with FreeBSD. The documentation is amazing, which is why things have gone pretty smoothly. And the system feels more engineered than slapped together.
 
kpa said:
PC-BSD wouldn't go anywhere if FreeBSD itself stopped offering half-baked (just my opinion) X11 based desktop environments that are barely usable for newcomers. PC-BSD could very well just take over the maintainance of the X11 and desktop environment related ports leaving FreeBSD in charge of the core technologies that are not X11 or DE related.
Click to expand...

On the other hand, PC-BSD will cease to exist if FreeBSD provides basic building blocks for a desktop right out of the box.
The Xorg server, drivers, fonts, xdm, TWM, Xterm, etc. From FreeBSD's own stable they just need to add a graphical package manager tool.
For beginners (at least for me, when I started using FreeBSD way back in 2001) setting these up are major stumbling blocks.
I believe if these are available out of the box from a standard installation then all these "Desktop" projects won't be needed.

To the topic of Systemd issue, which I believe got dragged into BSD world due to the road map (assuming there is such a thing :) ) Linux has taken towards graphical interface.

So, I'll throw my two cents.

Background and disclaimer.

I'm in IT for last 18 years. I work with a German engineering giant, household name there. Few years ago our communications business was carved out and merged with similar business from a Finnish (again a household name there) telecom giant. So you get the hint. I look after a product that we still maintain, enhance and provide to the telecom company that was borne out of the said merger.

So that establishes my technical background.

We use Windows as the desktop machine. All development is done there. All builds are made to pass on Windows, CentOS and FreeBSD. This is where I have a stake in this issue. People running development environments on either CentOS or FreeBSD do so on self initiative, without any IT support.

But wait, we have other lines of business (among few others like space and energy ) that do product development for media and transportation, where they use FreeBSD. Our in flight entertainment system and media system for cruise liners use FreeBSD.

I'm partial to FreeBSD myself, opted for Panasonic smart TV because of that.

First the systemd issue. Not many developers are aware of it. Folks that do architecture have just one concern, systemd seems to be aiming to control everything that exists between kernel and user land applications.
It is a single point of failure.
If it misbehaves you don't even get a half working system (something like Windows safe mode). You may need to reboot.

On technical ground other than this there is no issue. Systemd might do good to Linux world, it will stop the fragmentation.

Coming back to UI part, our own teams would have to take a decision if BSD get stuck in with a graphical stack that is in maintenance mode and is progressively less used by *NIX world.
For us X11 really doesn't matter. All we needs is a kernel mode driver and a pixel manipulation library, if X11 goes away. Simple because, the kind of applications our media and transport guys design always run in kiosk mode with some custom widgets. So strictly speaking no X11 is required.

However it will make it difficult for developers to run any kind of development environment on the target system.
Before someone jumps, I mean something like QtCreator.

So, with SystemD the following are the scenarios

1. Hell with SystemD - We don't need any GUI anyway.
2. Fall in line
3. Develop our own graphical stack, because we can't pay for the default BSD+GUI a.k.a Mac OS.
4. Fit Wayland to work without SystemD.
5. Maintain X11 forever.

For BSD world only #4 and #5 are the options.

Parting thought on #1.
There is no need for FreeBSD to spend money for KMS driver and plethora of ARM builds, if servers alone are the installation target.
On the other hand to be fair to the people asking for #1, no one is forcing the lay users to use FreeBSD as a desktop.
However, it is a blatant lie to say FreeBSD is server only OS (heck, even my TV runs on it).

But I think the truth as always lies somewhere in between.
Probably people that matter realize that the OS as a whole must be capable enough to run a graphical interface. Hence the work on KMS drivers, Wayland port and ARM portability (ARM ... certainly not servers).

Longevity of an OS is due to people using it, people writing applications for it and people keeping it up and running.
It is the third group of people that seems to vocal with #1.

Classic, so called "official" UNIXes must be have learnt this that hard way. For years we have fed with bullshit that critical infrastructure runs on so-called big iron.
For me it was an eye opener when one of the largest telecom providers in India (200 million plus subscribers) gave us firm directive ... only x86 Linux servers.

It is the conservative and vocal minority of the NIX world ... all of them BSDs, big irons, Linux ... combined, that failed to realize the value of decent UI. Hence the place for Wayland and unfortunately in the transition the SystemD guys took advantage of the melee to push their agenda.

Desktop computers are a dying breed anyway. Who are going to use them? Accountants with large spreadsheets, designers and us programmers. That's a very small percentage of total user base.

So, lets not lose sleep over SystemD.

All of our server applications are going to cloud and UI applications are moving to phones.
 
Hmm just little comment about desktop computers: There is one not so small niche group using desktop computers too. Pc-gamers. While people have predicted death of pc as gaming platform since at least Playstation 1, Pc gaming seems to be quite hard to kill. There are quite many game genres not decently available to consoles. And on other hand needing such amount of performance, that mobile devices are out of question. And you cannot simply install better GPU to laptop. So for serious gamer only real option is desktop pc. And quite many of them use quite substantial sums of money for their hobby.
 
PC gamers are an even smaller community.

"There are quite many game genres not decently available to consoles."

Its a matter of time these are available on gaming consoles or even Smart TVs.

For the gaming part, we should note that these "desktops" are used as custom made gaming consoles. They are no longer personal computers (I used the word desktop in that sense ... personal computers) in the sense we refer to them, which is the central theme of this thread.

In that sense, Desktop's are indeed a dying breed.

Its been quite a few years I've seen anybody buying desktops for home use. In the neighborhood also I've seen friends and acquaintances, opting for laptops as suitable for their budgets. Even our IT departments have stopped purchasing them. We are issued laptops (high end).

Gamers wouldn't be bothered by the issue we are discussing here, although they would keep alive a lucrative market for high end machines (no matter what runs in it).

BTW, there is this Gartner report that says PC shipment have dropped by 9.5% in 2015
http://www.gartner.com/newsroom/id/3090817

That's a steep fall in a single year.
 
SR_Ind said:
On the other hand, PC-BSD will cease to exist if FreeBSD provides basic building blocks for a desktop right out of the box.
Click to expand...
This statement seems to assume that you simply don't like what there is already. I understand that. Everybody has different tastes, but it still sounds like an insult when written here. For example I don't have any specialized Microsoft skills or familiarity so when I encounter a Windows machine, my first thought is that it is completely unusable as is and I would need to install and change so much that it simply wouldn't be worth the effort. That's me, and it's a matter of taste. It is not the fault of MS-Windows. To my view FreeBSD does (and does extremely well) already have the basic building blocks for a desktop right out of the box. Type pkg install x11/kde4 and there you are! If you don't like it that's not the fault of FreeBSD.

tldr; FreeBSD already has a desktop right out of the box. That box is called a "meta-port".
 
OJ said:
This statement seems to assume that you simply don't like what there is already. I understand that. Everybody has different tastes, but it still sounds like an insult when written here. For example I don't have any specialized Microsoft skills or familiarity so when I encounter a Windows machine, my first thought is that it is completely unusable as is and I would need to install and change so much that it simply wouldn't be worth the effort. That's me, and it's a matter of taste. It is not the fault of MS-Windows. To my view FreeBSD does (and does extremely well) already have the basic building blocks for a desktop right out of the box. Type pkg install x11/kde4 and there you are! If you don't like it that's not the fault of FreeBSD.

tldr; FreeBSD already has a desktop right out of the box. That box is called a "meta-port".
Click to expand...
Please read my post carefully. Right out of the box means whatever is provided via default install.

Please go through what I've listed that could be considered during default install - repeating for your perusal - Xorg server, fonts, drivers and xinit. I'm not asking for a desktop or even a window manager if you read carefully. An installation process can very well detect the hardware and install the appropriate display and input driver, copy some fonts etc. This much xorg stuff consumes negligible disk space.

I'm aware of what is KDE4. I have my own little desktop hacked out of various Qt based applications, many of them self written. Thank you very much.

There is something called "xorg-minimal" in the ports (you are aware of course) which can serve as a building block, if your intention was to educate me.

Since my comment was about basic bits to build a desktop, I don't understand how exactly meta ports becomes relevant here.

I might sound like an insult to you, but the oft claimed "server OS" does not live up to expectation of a server OS either from the standpoint of out of the box usability.
We have installation images for every obscure hardware out there, but the default installation won't give you something usable like a Web Server or a database server.
Sure, Apache or PostgreSQL may not be FreeBSD's software but they are not of Red Hat or CentOS either.

Every newbie that tries out FreeBSD (or any *NIX for that matter) is not a Windows or Mac OS point and click user of limited skill set.
Most people just don't have the time to go through pain of setting up a HTTP server or a database server after OS installation is done, when options are available.

As you can see we are not talking of GUI alone here, but the underlying malaise is same.

Lack of packaging.

It is not a criticism or any sort of value judgement, just a commentary on the prevalent tech culture.

>>

So, to the OP ... systemd or lack of it will not change anything.
We might end up with the most optimum and responsive Wayland implementation out there.
But first timers would still have to hunt around the ports directory to set up their desktops.
 
SR_Ind said:
Please read my post carefully. Right out of the box means whatever is provided via default install.
Click to expand...
I did. And it would appear that we have some difference of opinion. :) Of course you know well how to do this stuff. I'm probably much less skilled. However, my point is that there is no OS which works out of the box. Why single out FreeBSD as being significantly different.

Yesterday I spend several hours with a default MS-Windows install. It simply is not useful for basic file management and networking out of the box. I would have to spend many hours learning how to install appropriate software in order to do what I expect to do with the operating systems that I know (DOS, Linux, *BSD). From my perspective, FreeBSD only requires a few simple steps. Perhaps those steps are difficult for a person who comes from a MS-Windows background, but how is that different from my experience where what I need to do to make a Windows install usable seems daunting? (I say daunting, because I really didn't want to go there and just wanted to do a few basic things and then move on.) Windows is simply NOT ready for the desktop unless one has prior skills - one of which is familiarity with GUI installation and other MS-centric quirks such as not installing mouse and keyboard drivers by default. Again, I mention this because it is a fresh example of how even Windows requires a lot of work to use from the beginning. All available operating systems take time and special skills to get to the level that people like you and I are willing to live with.
 
OJ said:
I did. And it would appear that we have some difference of opinion. :) Of course you know well how to do this stuff. I'm probably much less skilled. However, my point is that there is no OS which works out of the box. Why single out FreeBSD as being significantly different.

Yesterday I spend several hours with a default MS-Windows install. It simply is not useful for basic file management and networking out of the box. I would have to spend many hours learning how to install appropriate software in order to do what I expect to do with the operating systems that I know (DOS, Linux, *BSD). From my perspective, FreeBSD only requires a few simple steps. Perhaps those steps are difficult for a person who comes from a MS-Windows background, but how is that different from my experience where what I need to do to make a Windows install usable seems daunting? (I say daunting, because I really didn't want to go there and just wanted to do a few basic things and then move on.) Windows is simply NOT ready for the desktop unless one has prior skills - one of which is familiarity with GUI installation and other MS-centric quirks such as not installing mouse and keyboard drivers by default. Again, I mention this because it is a fresh example of how even Windows requires a lot of work to use from the beginning. All available operating systems take time and special skills to get to the level that people like you and I are willing to live with.
Click to expand...

Lest we go off topic.

1. Windows out of the box - I'm not sure what version of Windows you are using. I've Windows 7 on my corporate laptop (Fujitsu) and Windows 8.1 on my home laptop (HP which dual boots - FreeBSD as the second OS). I never needed to install any driver on either of this. Windows explorer serves my purpose as a file manager. Internet explorer works by default on my home laptop. In my corporate laptop it requires me to adjust the proxy settings sometimes.

2. Windows for developers - Something like IIS server doesn't needs installation. You just need to tick it in the Windows component. Works the same way in Windows versions / editions on servers (where it is on by default) and desktops. Same with SQL Server, MS developer tools will offer to install a stripped down version. Developer tools are specialized tools, so understandably it requires a installation process. Not talking of third party software here.

However, this is an unfair comparison. MS owns all these products and they have the manpower to bundle all this together.

Apt comparison will be CentOS (or its commercial counterpart RHEL).
The installation process offers presets for "Desktop", "Minimal/Basic", "File Server", "Web Server" and "Database Server". IIRC recent versions of Debian too offers similar options. I'm talking of text /curses based installation program, not the GUI.

Even we confine ourselves strictly to BSD's, IIRC NetBSD (last used way back in 2010) has xorg in the base itself. They have even less funding and manpower.

How do they manage it?

If an installation process is able to land the user to something like TWM running with XTerm, he/she is halfway through with desktop preparation already.
 
Status
Not open for further replies.
Back
Top