Closed What Would You Like to See FreeBSD Do Differently?

Status
Not open for further replies.
I would like freeBSD to focus more on Desktop users. I know for long time FreeBSD was regarded as a Server material but if FreeBSD don't make it easier and the whole experience is smoother for people to use it on their laptops we will loose the next generation for Linux especially university students.
I recommend having a desktop tool library which would be installed with whatever desktop environment in ports. and this should have GUI package management tool, network tool, mount tool and etc...
 
Please do not misunderstand me. I am not trying to get on anybody's case.
Not to worry, that was just me using a common idiom. :)

The question was "what would you like to see FreeBSD do differently?" Since somebody asked, that was my two cents.

As a server, I don't see any problems with freebsd. But, as a desktop, it is a little surprising that I have to make special efforts to get freebsd do, what has been taken for granted in mac, windows, and linux, for the last 20+ years.

My point is that FreeBSD is basically a server OS, and the limited resources available are directed towards that purpose. It should not come as a surprise that developers do not abandon that ideal. Once there is a surfeit of developers, then there will be time for other things.

I am actually a little taken aback by your assumption that FreeBSD has comparable resources to Linux. Also, for your information both Microsoft and Apple are very large commercial companies with resources probably a million times greater than FreeBSD - probably even much more. Apple is the world's richest company - even richer than EXON. Little ol' FreeBSD is nothing in that world. :)
 
Not to worry, that was just me using a common idiom. :)



My point is that FreeBSD is basically a server OS, and the limited resources available are directed towards that purpose. It should not come as a surprise that developers do not abandon that ideal. Once there is a surfeit of developers, then there will be time for other things.

I am actually a little taken aback by your assumption that FreeBSD has comparable resources to Linux. Also, for your information both Microsoft and Apple are very large commercial companies with resources probably a million times greater than FreeBSD - probably even much more. Apple is the world's richest company - even richer than EXON. Little ol' FreeBSD is nothing in that world. :)

I totally agree with you... but for FreeBSD community to grow and in order to maintain FreeBSD position in the server side, it needs to recruit new people especially students. FreeBSD should make it easier and more fun for new people to dare to install it on their laptops and start playing with it and hopefully one day they will join the community and become contributors in all different capacities.
For this to happen a bit more investment in the desktop side is required. Big corporations these days outsource their DevOps work to companies like IBM and Infosis, etc... (I worked in Australia biggest Telcom and now in one the major 4 big banks as they call them here and pretty much this is what happens across the board) and these humongous corporations only support basically red-hat Linux for all the reasons you can think of. So for freeBSD and the like to survive they (we) need to penetrate the startups and academia sector (good examples whatsapp) to do so we need students and the new generation to get involved... which is means smoother desktop experience.
 
I wonder if it would be possible for a company to grow around FreeBSD the way that Red Hat has grown around Linux?

Red Hat started as an alternative to proprietary OSes. When Red Hat started, Linux was not very appealing to average desktop users.

Now that Red Hat is another Microsoft, maybe FreeBSD could be the next Linux?

Could be a great business opportunity. But it would be a shame if FreeBSD got ruined like Linux.
 
IMHO, what really made "Linux Desktop" become interesting (or more interesting) for some was Valve SteamOS, not because of the games or SteamOS it self, but because the several improvements that come together as result of its (Valve) influence, like the big improvement on Linux graphic stack had - specially on AMD/Radeon side.

If, hypothetically, Valve decide tomorrow FreeBSD would be a better platform for SteamOS instead Linux, and I think it would be (being SteamOS a serious project or not), FreeBSD would certainly have several desktop related improvements quite fast.
 
For this to happen a bit more investment in the desktop side is required. Big corporations these days outsource their DevOps work to companies like IBM and Infosis, etc... (I worked in Australia biggest Telcom and now in one the major 4 big banks as they call them here and pretty much this is what happens across the board) and these humongous corporations only support basically red-hat Linux for all the reasons you can think of. So for freeBSD and the like to survive they (we) need to penetrate the startups and academia sector (good examples whatsapp) to do so we need students and the new generation to get involved... which is means smoother desktop experience.

True, and the difficult part here is to convince academia that despite the fact that more and more companies expect people to have Linux-specific skills (as opposed to a generic, deeper understanding of how things actually work), offering to teach on FreeBSD has actual value on the market.
 
With all discussion about sane defaults on FreeBSD, while several users find the current defaults are the right ones (or preferable) and others think on something different should be used (servers, desktop, etc specifics), why not add profiles to the installer? For instance:

  1. FreeBSD Default
  2. Desktop
  3. High Secure Desktop
  4. Server
  5. High Secure Server
  6. ...
IMHO, it would simplify for most, and not harm the others.
 

And here is a default that always bugs me. Is it an absolute given that ALL desktops have to have an office suite? It's a very large package to force on people only to have to remove afterwards if not needed. This is not just FreeBSD, but it seems that all open source distribution do this.
 
And here is a default that always bugs me. Is it an absolute given that ALL desktops have to have an office suite? It's a very large package to force on people only to have to remove afterwards if not needed. This is not just FreeBSD, but it seems that all open source distribution do this.

I was meaning about things like sysctl.conf and security options (maybe firewall too), not third party software. I failed to explain that.
 
With all discussion about sane defaults on FreeBSD, while several users find the current defaults are the right ones (or preferable) and others think on something different should be used (servers, desktop, etc specifics), why not add profiles to the installer? For instance:

  1. FreeBSD Default
  2. Desktop
  3. High Secure Desktop
  4. Server
  5. High Secure Server
  6. ...
IMHO, it would simplify for most, and not harm the others.

I'm not convinced this would simplify things: I'm afraid this would lead to a more fragmented knowledge ("Sorry, I can't help you, I don't know how the settings are in your installation profile") and encourage users to skip reading the manuals and think they can be safe without understanding their system really. I might be exagerating a bit here, but this doesn't seem unlikely to me.

Having the same install as 99% of people (excluding the fully manual and fully automated installs) is a feature to me: I get the same base as everyone and I get the same docs as everyone to shape it as I want. I can follow the advice of those who know better. That wouldn't be impossible at all if we had installation profiles, but I think it would be harder.
 
Having the same install as 99% of people (excluding the fully manual and fully automated installs) is a feature to me: I get the same base as everyone and I get the same docs as everyone to shape it as I want. I can follow the advice of those who know better. That wouldn't be impossible at all if we had installation profiles, but I think it would be harder.

But 99% of the users change this same installation, so it differs from one another. I like the idea of having some feature sets like lebarondemerde suggests.
 
But 99% of the users change this same installation, so it differs from one another. I like the idea of having some feature sets like lebarondemerde suggests.

Yes, everyone changes it, but everyone starts with the same reference. I must say I do like the idea of automated configuration, but I'm just a bit cautious regarding the install itself. I'm thinking maybe the profiles could be applied via some kinds of metaports? This would leave a clear trace in the system regarding its configuration.
 
Oko commented on replacing Clang/LLVM in the base installation earlier, and suggesting lang/pcc.

I'll reiterate, that it would be good if in a future release that a lightweight compiler were included that can compile the kernel and FreeBSD base. GCC should be moved out completely. Then leave LLVM/Clang out of the base installation, except for a few libraries, and a program that asks, which Clang version would you like to install a package of or compile. Clang has to be installed a second time to a newer version for some programs.

It seems that, pcc can only compile C and not C++ or other base and kernel requirements, if I'm not mistaken. TenDRA (lang/TenDRA) however does take care of C++; it is no longer in ports, because the last website it had is no longer, but it's at http://www.tendra.org/. Its license is unclear, but it says current development is BSD licensed. If some lightweight combination of pcc or TenDRA can compile kernel and base, that would be cool.

I removed all compilers and other programs from src.conf to build a base, in an attempt to use a package of devel/llvm39. It did shave 2 hours off of base compile time, but it left me with a system that could only use packages or could only compile basic C programs, but not the kernel.

Another suggestion is for base compilers' and their tools' directories to be under one directory such as /usr/compilers/. I'm not certain, but it seems that newer versions from packages of clang rely on parts from the base installation, which has a duplicate replacement.
 
TenDRA (lang/TenDRA) however does take care of C++; it is no longer in ports, because the last website it had is no longer, but it's at http://www.tendra.org/. Its license is unclear, but it says current development is BSD licensed. If some lightweight combination of pcc or TenDRA can compile kernel and base, that would be cool.

You do realize that the deprecated message on Tendra was written in 2011, right?
It is accurate to say it was removed in August 2011 because nobody addressed the missing upstream, but 6 years later I would say it's straight-up misleading to mention the website now. And obviously you can't just revive the port because 6 years of stagnation guarantees it is not buildable.

Bringing in Tendra would be a new effort -- like any new port.

I'm not going to comment on the actual concept of replacing the base compiler -- but I will point out that OpenBSD considered PCC to the point of bringing it into base, and then later removed it. The conclusion is that it wasn't feasible to use.
 
I would have thanked your post, for being informative, except your misunderstanding, but you want to call my post misleading? I do realize that project was depreciated, and the point was, its old website, which is no longer as pointed out by the Freshports page, and must be years old, has been replaced by a new website, and is more recent than that.
 
Again, what you are saying is irrelevant.
Freshports keeps data on deleted ports. You actually have to select "deleted ports" to get it to show in the search results.
Obviously, after a port is deleted, the freshports entry is never updated again.

Given that, how is it relevant what Freshports "points out" now?
How is it relevant that somebody started a new domain that makes it seem that Tendra is maintained?

I don't see the relevance.
The port is dead, dead, dead.
It cannot be revived without a complete rebuild.

You might as well pick any software not in ports, and then point out a current website. It's really the same thing. So yes, talking about lang/tendra is misleading because you make it sound like the problem is trivial and of course it's anything but. It would be more honest (for lack of a better word) to say it should be added to ports (as if for the first time).
 
Your point is ridiculous. You act like you dont know Freshports has deleted ports, then you act like, it has deleted ports. Stop flipflopping.

Your original point is inaccurate.
It's not "no longer in ports because the website is gone", it's no longer in ports because somebody removed it way back in 2011.
The reappearance of the website is irrelevant.
 
I said, it is no longer in ports. Just like for the reason given in Freshports. If you are upset that the ports tag can be used on deleted ports, don't look at me, because I said, "it is no longer in ports." I don't know how many ways that can be interpreted or misinterpreted.

My point was, which I didn't directly say was. You can use pcc, then just adopt, implement, model, or copy parts from Tendra, that are allowed. Isn't that possible, without porting it? Where I said some combination of.
 
If your point is "Tendra used to be in ports but you can use it to compile kernel and base if you go its new website" then my point still stands.
Which is mentioning the reason it was removed from ports is misleading and irrelevant. You could have removed "because the last website it had is no longer" and not only would your intent remain intact, it would actually have been clearer. You muddled everything up by focusing on the unimportant reason for removal 6 years ago.

Finally: Where's the proof that A) Tendra currently runs on FreeBSD and B) that it will actually compile the kernel and/or the base?
 
Status
Not open for further replies.
Back
Top