Will Mir Come On The Ubuntu 14.04 Desktop?

[blackout23]

C is best choice only for developers who aren't smart enough to learn C++. It is also possible to use assembler with C++ and create kernel in C++.
http://www.osdever.net/tutorials/pdf/cpp_kernel.pdf

http://harmful.cat-v.org/software/c++/linus

If Torvalds says C++ is horrible there is probably some truth to it. He has more knowledge and achived more than all of the users on phoronix.com combined.

[bkor]

C is best choice only for developers who aren't smart enough to learn C++. It is also possible to use assembler with C++ and create kernel in C++.
http://www.osdever.net/tutorials/pdf/cpp_kernel.pdf

You might want to read up on something called Linux. Also a kernel.

[matzipan]

Modern distros are perfectly capable for home usage (web, email, music, videos, office, etc.). As a matter of fact, I found that both unity and kde versions of 13.04 to have finally reached production quality. Now what it needed is modern ui themes for unity and kde, in order to catch up with recent developments in android, ios, windows, and hardware deals. I see HP selling Chrome laptops, which IMO is nonsense, but not ubuntu. Why?

I think you're kidding. Modern distros are capable of home usage IF you're using Intel and not AMD, IF you're using Intel not nVidia or ATI, IF you're not using broadcom, IF you don't have switcheable graphics, you get the drill... just too many IFs. Oh, and installing any modern distro on any computer that has UEFI and Windows 8 preinstalled? A pain. They work perfectly fine in the corporate or education area were you can make the hardware tailored to your solution, but not in the consumer area, especially not in a world dominated by UEFI and OEM W8.

If thats the case, than Ubuntu is just pointless. It soon (2-3 years) will be just another obsolete/crappy distro, we should encourage people to switch to other, better distros for desktop than.

I don't see any other distro getting momentum in the consumer linux world. Elementary is still something for the enthusiasts. Maybe Linux Mint, but they lack the corporate backing really. Or Fedora... If Red Hat ever thought Desktop Linux is actually a good idea.

[GreatEmerald]

Considering they postponed Mir deployment for a year, it's pretty funny to think about all the people who were proud how Mir would be deployed in tier-one distributions before Wayland (despite that not actually being relevant), now that the opposite is true.

[r_a_trip]

I don't see any other distro getting momentum in the consumer linux world.

So you are saying our best bet for desktop Linux is a company who is turning the desktop into a gimmicky addon for mobile phones.

We're screwed.

[silix]

If Torvalds says C++ is horrible there is probably some truth to it.

yes, that he doesnt like C++ and that he does without it as far as his coding needs go

He has more knowledge and achived more than all of the users on phoronix.com combined.

judge someone's words by themselves, not by his alleged authority...
hate me if you want, but do yourself a favour: start with the most pragmatic mindset you can, assume that developers that dont limit themselves to one language and can produce masterpieces of simplicity and efficiency using C++, do exist (i know some of them, one is my best friend and the one person i envy), and reread that disregarding the name

You might want to read up on something called Linux. Also a kernel.

the linux kernel was born as a unix clone - to leverage code already available for unix at the time, and through this, contribute to and perpetuate the unix culture
thus, writing the kernel in the same language the original unix kernel and most of the userland was written, was only logical - also because, the C compiler is the first (sometimes the only one) compiler available for several platforms (ie a lowest common denominator) thus the only choice for building a portable kernel
moreover, minix on whose code Torvalds based his own kernel at the very beginning, was C based
(if it hadnt been for these factors, who knows, Torvalds -whose goal was simply to make a free operating system for his own needs - may have gone for a VMS clone written in Ada.. or again maybe not because VMS didnt have the same kind of hacker culture associated with it... but again, who knows)

the one thing Linux being written in C does NOT tell, is C is the only possible language for kernel development - inferring so is a fallacy

[talvik]

"So I will bring it to the always insightful and controversial forum goers, do you think Mir will make it to the Ubuntu 14.04 desktop by default?"

Always insightful?

They shouldn't put Mir/XMir/Unity 8 on 14.04LTS, but they will. They are in a mad rush to push their vision, being LTS didn't stop then in 12.04 it won't stop in 14.04.

[mrugiero]

Newbies can be minor contributors if there is proper code review, but major ones should be experienced. Newbies can get experience also with less important software.

They still need to get to know the code base.
IMO, C++ used right leads to simpler code, which means newbies get to know the code base faster. It just requires they learn C++, which is harder to learn than C. But afterwards, reading C++ becomes easier than reading C, I think.

C is best choice only for developers who aren't smart enough to learn C++. It is also possible to use assembler with C++ and create kernel in C++.
http://www.osdever.net/tutorials/pdf/cpp_kernel.pdf

So? Being able and it being a good choice aren't the same thing. Nobody said it was impossible.
Until not too long ago, it was a failure for low level because of unneeded copies. C++11 changed that, but it still might have some problems of the like for low level.

Note I'm not pro nor against it using C++ in low level, I'm just pointing out there are flaws in your arguments. I don't point out the ones of the other user because you already did, but I acknowledge them, too.

yes, that he doesnt like C++ and that he does without it as far as his coding needs go
judge someone's words by themselves, not by his alleged authority...
hate me if you want, but do yourself a favour: start with the most pragmatic mindset you can, assume that developers that dont limit themselves to one language and can produce masterpieces of simplicity and efficiency using C++, do exist (i know some of them, one is my best friend and the one person i envy), and reread that disregarding the name

Yet, the only example you gave proves nothing on the efficiency camp. On simplicity, you are right.

the linux kernel was born as a unix clone - to leverage code already available for unix at the time, and through this, contribute to and perpetuate the unix culture
thus, writing the kernel in the same language the original unix kernel and most of the userland was written, was only logical - also because, the C compiler is the first (sometimes the only one) compiler available for several platforms (ie a lowest common denominator) thus the only choice for building a portable kernel
moreover, minix on whose code Torvalds based his own kernel at the very beginning, was C based
(if it hadnt been for these factors, who knows, Torvalds -whose goal was simply to make a free operating system for his own needs - may have gone for a VMS clone written in Ada.. or again maybe not because VMS didnt have the same kind of hacker culture associated with it... but again, who knows)

IIRC, Torvalds didn't base on Minix code, since that was not allowed at the time. He based his learning on it, but didn't use the code. You are right in everything else, AFAIK.

the one thing Linux being written in C does NOT tell, is C is the only possible language for kernel development - inferring so is a fallacy

True.

"So I will bring it to the always insightful and controversial forum goers, do you think Mir will make it to the Ubuntu 14.04 desktop by default?"

Always insightful?

It probably sounded aggressive to write "always controversial and from time to time even insightful" :P

[ciplogic]

They still need to get to know the code base.
IMO, C++ used right leads to simpler code, which means newbies get to know the code base faster. It just requires they learn C++, which is harder to learn than C. But afterwards, reading C++ becomes easier than reading C, I think.

So? Being able and it being a good choice aren't the same thing. Nobody said it was impossible.
Until not too long ago, it was a failure for low level because of unneeded copies. C++11 changed that, but it still might have some problems of the like for low level.

Note I'm not pro nor against it using C++ in low level, I'm just pointing out there are flaws in your arguments. I don't point out the ones of the other user because you already did, but I acknowledge them, too.

Yet, the only example you gave proves nothing on the efficiency camp. On simplicity, you are right.

I think the arguments that C or C++ is faster is silly in its own right, but in general I think people should expect a tad faster writing things in C++, and this is shown also by the coding of the performance driven codebase are written a lot of projects:
- most of big game engines are C++
- many libraries of high performance are C++

But even so, the reason why C++ is good for infrastructure is simply because C++ improved. Not necessarily just with std::move, or with better compilers, but it is good enough to execute better than most other constructs why keeping safety of a big system. Proof to that is at multiple levels, but one part which I want to stress is from C++ creator and his 2012 page was considered a notable paper of 2012 and one of the best of ACM 2012 papers.

The fact that you can write the code in a "C-like", "zero abstraction" like manner, is already abstracted for you in many STL algorithms. But most people don't do it with zero abstraction, either in C or in C++, so more code is referenced and is written in C++,the most likely is possible to optimize it more, or to be more robust.

"Infrastructure software needs more stringent correctness, reliability, efficiency, and maintainability requirements than nonessential applications. " - Bjarne Stroustrup

[dee.]

C is best choice only for developers who aren't smart enough to learn C++. It is also possible to use assembler with C++ and create kernel in C++.
http://www.osdever.net/tutorials/pdf/cpp_kernel.pdf

It's also possible to pick your nose with your dick, but that doesn't mean it's necessarily a good idea.