my little rant...

[Xen]

Colin Watson schreef op 11-11-2017 0:18:
> On Fri, Nov 10, 2017 at 02:03:18PM +0100, Xen wrote:
>> I'm still on 16.04. I still want to use kernel 4.6.
>> 
>> It's not available, we're on 4.10 now and 4.13 coming on.
> 
> Each kernel version we offer has to receive security support.  We have
> teams of people doing this, and it takes real time, energy, and money,
> all of which are limited resources.  It simply isn't viable to support
> more than a small number of kernel versions at a time on any given
> release.

I was going to compile it myself but just couldn't get around to it, my 
mistake I guess.

Also I don't know how feasible it is to do that with today's 
infrastructure.

But for my own feeling I still need to complete that step; I have 
skipped a step as it were ;-).

But you know, this is not directed at the kernel.

And I appreciate your efforts. I have seen the huge amount of commits 
passing through the ubuntu kernel mailing list.

I also know... I mean.

Never mind that.

The point in general I would like to make at some point is that I wish 
that some developers, not necessarily you or Ubuntu's, would also take 
into consideration the real time, energy and money their lack of effort 
to make something completed, costs their users.

I know in general Ubuntu has put a huge amount of effort into finishing 
and polishing Unity, and Unity is not KDE.

But I also feel the Kernel itself is rushing ahead too fast.

Not just for me. In general.

Ubuntu now introduced...

well whatever right.

You can keep changing infrastructure down the hood not in terms of UI 
but the basic system...

sponsored by some efforts by some other linux supplier...

I just think we are still in a rush to nowhere.

All of the stuff that has been changed and that is continued to be 
pushed.

Is not done.

Will take years to complete it and people will at some point have to 
stop what they're doing and revisit what they haven't...

What still needs to be made right.

They will have to stop the push, and go back to basically what is 
currently 16.04(.0)

And look at what isn't right and I'm talking systemd mostly.


Not trying to sound ungrateful here for people doing work.

But sometimes 60% of that work is due to _doing too much_ of the wrong 
stuff.

If you put less hay on your fork, you also have less work to do about it 
you know.

And I think that in general the lack of resources in Linux comes down to 
creating too big a burden for yourself.




>> Whenever a version of (K)ubuntu is released the devs don't even wait 
>> to look
>> at what they have done, they immediately rush on to the next thing.
> 
> It's always going to be the case that some bugs remain unfixed and 
> users
> are unsatisfied, pretty much no matter what, and I don't necessarily
> think that our post-release support has always been everything it could
> be either.  But really quite a lot of time and effort goes into issuing
> updates to Ubuntu stable releases.

The point is more that you have to stop to consider what you've done 
before you rush on, or you can never know whether what you are going to 
do next is going to be the right thing.

I'm sorry for sounding ungrateful but I'm also quite sure 90% of those 
updates are security fixes and nothing structural or perhaps I wanted to 
say tangential, of course it couldn't, but I'm just talking about 
reflection.

I haven't been following any Ubuntu newsletter myself and the Ubuntu 
Weekly is hard to read, or I could say hardly readable because it is 
full of links, just full of links and no proze.

However the Fridge at 
http://fridge.ubuntu.com/2017/10/04/ubuntu-weekly-newsletter-issue-520/ 
is an accessible resource.



Regardless it always appears to me that the focus is purely with 
enterprise and Ubuntu's main focal point is containers. I'm not saying 
that's a bad thing. I'm saying it's apparently a niche that was well 
received and in which Ubuntu has been able to specialize.

What results is a framework for automation that ordinary users can't 
access.

I'm just a home user at present. I dabble with containers, but limited, 
and only LXC.

In a world where people never solved the basic linker problem (often 
having fixed library link paths if you compile something) we now have 
containers that solved the problem entirely but are useless for the 
ordinary person.

So while for the enterprise many things are solved, ordinary, old, 
"regular" Linux systems are left behind.

There is a frenzy to automate deployment of workloads but...

The basics in the old, regular Linux system haven't even been covered 
yet.

I can't just take an application and put it on USB stick.

That simple use case is not covered.

I can't take anything out of the FHS and just transplant it into its own 
"container".

So where are the basics?

Suppose there was a world where every other person had a Linux system or 
a system capable of running Linux applications.


Suppose there was a modicum standard way of running GUI applications, 
such as when Windows would have native (XMing) support, or perhaps 
better, if it was Wayland(?)




Suppose every Windows PC, or every Linux PC for that matter, could run a 
form of containerized application the way Snaps were really 
envisioned.... but without anything special like Snaps?

(Which is a huge misnomer to begin with but anyway)

How am I going to take an ordinary package.

Wrap it up including its dependencies.

Define if we must a necessary Core as you have done.

But in general just take an existing package, source all dependencies, 
and put them into either one container format or one hierarchy tree 
(which would at this point basically be the same thing)

Put them on USB stick.

And I can run them on another computer?

Not much different from Java jars.

But it doesn't exist, does it.

In Windows I can download PortableApps and I'm done.

I can put it on an encrypted USB stick and I'm done.

Mobility achieved.



Rolling releases is not a very good idea but anyway. I think it causes 
library dependency creep.

You can't really put structural changes to a system in a rolling release 
anyway, can you?

So where is the basic functionality for users that Windows people have 
had for decades, basically?