Re: Proposed documentation, please comment! [was Re: Bug#838919: debian-installer: please calculate swap parition according to max RAM...]
Wouter Verhelst schreef op 22-10-2016 21:39:
On Sat, Oct 22, 2016 at 02:43:13PM +0200, Bart Schouten wrote:
I have no problem with what Debian is today,
Actually, you do. You seem to be advocating against systemd; but
systemd
*is* what Debian is today. Jessie has been released with systemd as its
default init system, and there are no plans to change that.
Why the hell are you bringing systemd into this discussion. This was
about documentation foremost.
I am not arguing against something that exists today, I am arguing
against
trying to turn it into a user-friendly system for people like my
little
sister, which I don't have, but she is 34 and would never be able to
use
Debian (or any other Linux variant) to any reasonable degree of
proficiency,
That's your hypothetical experience. I, on the other hand, know
millions
of hypothetical lay people who are pefectly capable of using Debian.
Why don't you try to insult me more visibly, and insult everyone's
intelligence at the same time.
You claim I want to move Debian backwards.
You are placing words in my mouth.
You say my opinion comes down to making Debian difficult. I interpret
that as a backward motion compared to your wanting to move Debian
forward in becoming easier.
However, I hardly do a thing other than solving my own issues and
sometimes writing documentation about that.
How that can "make" Debian "difficult" is beyond me. I would first need
to perform some magical action to turn Debian difficult, now wouldn't I.
Not touching it can hardly render something difficult. You attribute
some kind of magical power to me.
A stone you don't touch is not getting moved.
A Debian I don't touch can also not become difficult because of my
actions.
If I only advocate against something the only consequence of that is the
status quo is preserved. If I turn off the tap, the water stops running,
and the basin doesn't overflow.
I am only saying: don't do something. That can hardly cause Debian to
become difficult.
No no, I am not doing that. I really am talking about "making something
difficult", which is a good way to avoid getting more users.
You'll still need to explain my magical powers in achieving that. You're
attributing nonsense to people here. You have a political agenda, I can
see that, but I'm not part of that.
Neither do I. Systemd actually saved me a run to the data center
once[1]. That's a *good* thing.
What the hell are you talking about. We were speaking of documentation
and taking people by the hand in achieving stuff they otherwise wouldn't
have.
The topic was the swap partition and what size it should have. The
proposal was to write a step-by-step guide for people to know how to
resize their swap partition to keep up with a changed memory (RAM)
footprint. In response to that I said that I thought Debian was not for
lay users, and you responded to that, and to nothing else.
In the context of that statement I uttered that I felt "take by the
hand" documentation would not be good for Debian in the sense that you
would need constant effort to provide such documentation as a greater
power is to be had in allowing people to understand what is necessary
themselves and then for them to do that.
I said these things in the context of the proposal to write some
documentation that would allow a novice user to understand or at least
to perform the steps required to increase the size of the swap
partition, to which I offered some help, but I also noted that it would
be a bit silly (in those terms) to think that e.g. Windows users who
know nothing about computers should be using Debian in the first place.
I think that is just not realistic to begin with.
I frequently help people on the grub user mailing list to repartition
their drives. At least, the same question keeps popping up.
But those people display a lot of technical knowledge and still don't
know how to do it (because some piece of the puzzle is missing,
usually).
Ordinary people generally do not sit on computer mailing lists and if
anything they just ask questions on a forum. But a non-technical user
won't be able to do even that.
Your million hypothetical lay users are not non-technical people. People
are extremely daunted when they even see those instructions. So I don't
know what you are talking about. You have a misconception of what "lay"
users then are, and how much technical you can expect of people in
general. I know my sister, you don't. I know how much she panicked with
Windows itself. I know what would happen if I wasn't there for Linux.
You can't expect the general population to use a Linux system, you just
can't.
What Ian said was that you can expect a lay user to be able to use Linux
(or Debian) if he has an expert nearby. I agree with that. You say I
don't. Whatever. That still doesn't make Debian suitable for those Lay
users (without the experts, nearby).
Again, you are placing words in my mouth.
"Making difficult" is an activity.
I am performing no such activity and certainly do not endeavour to move
Debian in some kind of "more difficult" direction by wanting to keep the
documentation at the level it is today, with respect to some proposal to
create "take by the hand tutorials".
I certainly enjoy creating documentation and I think it can be improved
overall also on the wiki but the goal of doing that to widen the scope
is a bad goal. If your documentation improves and as a result of that
the scope widens, that's fine. But you should write documentation for
its own sake in my opinion and not because you can then hope to attract
more users and then achieve some other goal such as more people
developing for it. The documentation should be the goal, not something
extraneous to that. Otherwise you get what's on the Ubuntu wiki, with
all due respect, where documentation is often politically motivated
(from the viewpoint of Canonical, I assume).
The documentation itself should be the goal, not something else. But I
will stop writing again here, I have said enough.
My position is simply this:
* focus in documentation on users understanding the system rather than
being taken along in something they don't understand.
* create architectural documentation and overview documentation that
allows users to understand what they are dealing with
* allow users to make their own choices with regards to that and don't
force them into paths they don't understand or with the promise that if
they take step by step steps that then everything will be well when you
can't actually deliver on when it goes wrong. People need to understand
for themselves what they are doing, you can't be there when disaster
strikes.
The promise I think we should make is that Debian can be used by people
of all kinds of backgrounds -- experts and lay people alike. The former
will probably customize their system; the latter won't. That simply
means we just need to make the defaults work really really well.
You are simply copying Ian's words here. This topic was about
documentation (see the title). I don't buy into this "lay users will
never want to customize their system". The problem with those defaults
is that they will often break to moment you change *anything* in the
current way of doing things.
I do not oppose a workable default system, why should I. But the problem
is that if users change a *thing* suddenly things don't work anymore and
everyone wants to change something, for the most part. But the system is
not very forgiving.
It *barely* works, is what I mean, if you go down that path.
Example:
* LVM on disk that has RAID signature from BIOS raid tool is not getting
recognised by libblkid
* udev rules for LVM depend on libblkid to recognise PV for automatic
activation, and as a consequence these disks do not get activated
* was fixed in recent util-linux at my request/suggestion/petition
Actually I believe they fixed a different issue (for me) and RAID
signature still takes precedence. That means that anybody who uses BIOS
to create raid signature and then actually doesn't use it, and then
creates bootable Linux system with LVM on same disk, will have a
dysfunctioning system that just won't boot/work.
Prior to systemd as a matter of fact this problem didn't exist because
the activation didn't depend on udev and was simply a call to vgchange
-ay which *does* activate everything including the "corrupt" disks.
So we have seen a regression here and the system just *barely* works.
The smallest thing can throw it off. That is not good design.
"Lay users will never come across that" is often cited as a reason for
bad design. "People won't see it anyway". We mustn't buy into this idea
that there is a kind of "novice desktop user" who will be okay with all
the defaults and that we can sell this system to the user on the premise
that as long as this user sticks to the defaults, it should work. That
is very very fragile. Breackage is imminent.
That (making the defaults work) is a promise we can easily keep.
Debian is just not a very end-user oriented system. Set your goals
accordingly. I certainly do not look to Debian to provide my desktop.
Ubuntu has a KDE team that focuses exclusively on that, it is called
Kubuntu. Why do double work, if you are part of the same ecosystem.
I use Debian for servers, it is my preferred server environment. I do
not really consider anything else, actually. Apart from perhaps FreeBSD
down the road. I don't mind the system being solid, why do you think I
have a problem with that. I was talking about documentation only here,
mostly, until you brought up that other part.
[1] kernel OOM'd sshd. systemd restarted it.
Such an achievement in electrical and digital engineering. I think we
can make our way to become part of the galactic civilisation now, with
these advances.
Reply to: