Re: Desktop NetBSD needs your help

[Aleksej Saushev <asau%inbox.ru@localhost>]

Dima Veselov <kab00m%lich.phys.spbu.ru@localhost> writes:

> On Sat, Feb 07, 2009 at 11:59:53PM +0300, Aleksej Saushev wrote:
>
>> > I don't think NetBSD is appropriate for novice user desktop use, and I
>> > see that as about as much of a problem as it is that my hacksaw isn't
>> > appropriate for driving screws.  The world has plenty of screwdrivers;
>> > I'd much prefer NetBSD focus on making a good hacksaw (largely because
>> > I don't see anyone else making hacksaws nearly as good).
>> 
>> How do you expect new users to appear then? Do you think that NetBSD
>> users are born experts?
>
> Reading such, I always wonder - what make people to use NetBSD and
> suffer from it's technical orientation. Tell me - why not just use any
> other user-oriented system and leave at least one system for people,
> who count blocks better than megabytes?
>
> NetBSD is really designed for people, moved to NetBSD by their wise
> looking for really working things, proved for simplicity. If you can't
> understand some deeply technical things - don't even try and use
> FreeBSD or Linux.

Technical design doesn't nullify need for aesthetic design.
Andrej Tupolev (and Oleg Antonov after him) considered aesthetic design
as one of primary criterium of success.

The whole point of the desktop subproject is to simplify NetBSD for
entry level user. Note, that noone says that that user is stupid,
he isn't that familiar with the system (its internal structure) yet.
The latter unfamiliarity shouldn't prevent user from doing useful job
at unrelated abstraction level.

>> In particular, do you expect that newcomer learns the Guide by heart?
>> Is fdisk/bsdlabel documented that well, so that it can be learnt by
>> heart before installation? Is interface that good?
>
> They definitely should know how computer generally works. You know -
> I believe NetBSD installer is the best installer in the world. Why?
> It is so much straight, blazingly fast, small, simple. All core NetBSD
> utilities is at most simple as they should be for people understanding
> how it works in production, not in a desktop play.

Half of them are simple to being useless.

PERL (very complex and very horrible tool) was written not "just for fun",
it was written because shell and AWK are pretty hard to use technically,
and so are LEX and YACC.

Nowadays part of PERL domain is taken by Python, another tool, which is
still more complex than NetBSD utilities. Still users, who do practical
work, install PERL or Python anyway. Because bare system is not enough.
This is so different from what it was in 80's.

> You wish a nice desktop or simple installer for your play with that
> wild unknown OS, meanwhile others use hundreds of NetBSD boxes,
> praying for noone coming with 'fresh' Linux-like ideas to turn it
> user-friendly. You just try to get a Volvo panel in a heavy-duty dump
> truck - believe, there is no choice, trucks wouldn't and shouldn't
> change their ugly look ever. They will grow itself, enforce wheels and
> enhance capacity, always charging driver to deeply understand what is
> going on.
>
> After getting driving license you can drive any car, but you wish to
> drive a helicopter. Write your own installer for NetBSD, include the
> distribution in it and be happy.

Wrong analogy.

Not so long ago, if you wanted tunnel microscope in your laboratory,
you had to build it yourself. Sure, it isn't that hard, if you know
what to do and are experienced enough. You had to do it anyway, since
the whole problem domain had just emerged and everything was in flux.
Building tunnel microscope nowadays is just a waste of time and effort,
because it is easy to get ready one, if all you need is using the tool.

Noone builds his own tools, unless they are unique or one learns how to
build those tools.

What you defend is:
a) one is to build one's own tools even when there's no need to;
b) one is to live in ugly looking lab with ugly looking tools
even when it isn't hard to make everything look better;
c) one is to do manual work even when there's no need to do so, since it
could be automatized long ago;
d) one is to live with primitive tools even when they can't deal with most
of problems, which became more complex than in past.

And all this is done under false premises that:
a) everything connected to Linux is bad and evil;
b) everyone around wants to cut down possibility to fall in "expert mode";
c) everyone around wants to make "dummy mode" the default;
d) truck has to look like "Kommunar" of 20's or "Stalinets" of 30's
to be functional at all, and its functionality should be similar
if not the same.

>> > Of course, like everyone else (well, except core/board/etc), I don't
>> > get to dictate which way NetBSD goes except to the extent I take it
>> > there.  This, I think, is largely why most of my comments about
>> > NetBSD's direction have been negative, because they've all been in
>> > response to people who want to change things, and what I think we need
>> > to do is to stop trying to change our direction and concentrate on
>> > doing well what we do well.  If you want a desktop system, you know
>> > where to find it - and I use that wording despite, not because of, the
>> > derogatory tone the original carried in its context.  (At work, I had
>> > occasion to set up a GUI desktop, and used FreeBSD; some would see this
>> > as a failing of NetBSD, but I don't, any more than I see it as a
>> > failing of my hacksaw that I don't reach for it to drive screws.)
>> 
>> Why do you resist changes? The world has changed since 90's.
>> Some things proved to be inefficient, even dangerous.
>> Do you think that we should leave inefficient ones as is
>> just because changes don't make you happy at once?
>
> der Mouse was 200% right. If you need functionality - bring it by yourself,
> leave developers on their track. Things you call 'proved for inefficient'
> were really proved somewhere outside NetBSD team and users, I don't know
> any inefficient mechanism in NetBSD. Some talk about it, they call such
> a names, NetBSD is old, inefficient, anything else. These people turned
> Linux in a bunch of visual bugs, where you cannot understand what is going
> wrong, when needed something unusual, and then have to post an error to
> google - if there one more guy got it. I have a really big collection of
> 'user-friendly' modifications, which were really destructable for the
> system, because covered real things under ugly interface.

If you prefer to live eyes closed and don't know any case of inefficiency
in NetBSD, it doesn't really mean that it isn't there. You don't need to
search through the Web to find it, just go through manual pages and read
respective caveats and bugs sections.

>> One of NetBSD advantages noted by newcomers is its high-quality
>> documentation, and it is documentation quality what can be checked
>> by novice users first. Without good documentation you can't tell,
>> why NetBSD is better than Linux.
>
> Yes, it have high-quality documentation. Using NetBSD I used documentation
> many times and everytime found an answer. It is always that sort of answer
> I need - low-level to understand what really happen.

Using the same documentation I found bugs in it and know several topics,
where documentation is simply missing.

In addition to documentation defects I know several bugs in some of those
simple utilities, which are pretty critical to common user. For instance
NetBSD pax can't handle files with long names, contrary to "Linux" GNU tar,
which handles them pretty fine.


-- 
HE CE3OH...