Drawing graphics on terminal

[Paul Robinson]

You've got me going. You've just touched on my favourite subject. Apologies 
for those of you who prefer short e-mails.

On Mon, Jun 16, 2003 at 03:18:52PM -0400, Leo Bicknell wrote:

> Some of this could be done in the current installer, if there wasn't
> an effort to make it still fit on a floppy.  Mind you, I'd like to see
> the floppy based install stick around for a while, but I think FreeBSD
> needs to embrace the CD reality.

Actually the current effort is useless without at least two floppies. but I 
get your point. The installer is something that has been bugging me for a 
long, long, long time now. I hate sysinstall. I had some spare time and was 
going to start on something when Jordan piped up with libh. I'm not sure if 
libh was the right way to go anyway - it just prettied up sysinstall and 
made it more confusing to a novice user. The issue however has continued 
clawing at the back of my head.
 
For me at least, the issue has raised lots of other important issues:

- The project will never agree what is required of the installer, as 
everybody has their own agenda. Some see FBSD as a server OS that could do 
desktop, others want to push it forward as generic "one-fits-all" OS that's 
as good at desktop work as it is at server work. The installer needs to be 
able to mirror the user's wishes there and then.

- Saying "CDs are the way of the future and we must embrace" is not only
short-sighted on the long-term game plan, but negates the alternatives - for
example, would it not be relatively easy to produce a single floppy boot
disk that grabs the installer off the network? Off an FTP site? On the
mirror? With the network config in a file that could be edited on the disk 
in notepad or vi? Why not? Why not *try* and make it work?

- We need to abstract the package management and /usr/ports more for the 
average user. You mother would not be able to able to build a secure web 
server on FBSD without a great deal of hand holding. Until she is, the 
project should realise that we're not in any way proponents of a "user 
friendly" OS. It's a sysadmin-friendly OS. Sysadmins would like user 
friendlieness just like everybody else, they just don't want the ability to 
stop things happening they don't want. Aim for making a system that would 
make a great sysadmin out of our mother, and the sysadmins will be 
considered gods. They will thank you, and repay you in patches, beer and 
possibly their own employer's sponsorship and employment.

- Whether the installer is graphical or not is not the issue. Grey boxes on
a blue background with yellow, red and black text is just plain ugly to a
society that understands art and interior design. I know you're limited on
pallet due to the restrictions of the console, but you can make sysinstall
nicer just by changing the colour scheme. You can make it a hell of a lot 
nicer by making it consistent and functionally useful.

- A graphic installer would be nice though, because novice users need a bit 
of cuddling in those first few scary hours when new to the OS. We should not 
be scared of cuddling out users. For FBSD to attract money and R&D, we 
should not pretend that this is a members-only boys club and if you don't 
know what a disk sector is you're obviously too stupid to get any benefit 
out of running a Unix-like operating system.

- The KDE and Gnome stuff going on around FBSD, in my opinion, needs a 
helping hand or three so that when somebody installs KDE on FBSD, there is 
FBSD related stuff in the menus, FBSD themes, the whole lot, a la Redhat, 
Mandrake, and so on. This is not dumbing down. This is helping people who 
want to run a fancy window manager get the most out of their system. It 
helps advocacy. And besides, those guys could do with the help anyway, it's 
a lonely thankless job, and yet it has one of the biggest potential impacts 
on manager and/or investor perception.

Anyway, there's loads more I could go on with, but you get the idea. The
installer is the least visible piece of software in the whole project in
general day-to-day use - which is worrying, because everybody would agree
that title should really be going to send-pr and no other - but it's also
the one that carries the greatest benefit to ports, pkg management, general
user experience, and so on, just by sitting down and thinking about it
collectively.

My vote is we should be looking to mirror the Solaris installer - when
you're installing single machines you can do other things in the background,
and when you're installing a machine room full of kit, JumpStart helps out. 
And yes, I am starting to work on this. It's not going to get even near 
Alpha before the year end though because I'm busy, but if somebody wants to 
grab the design out of my hands, drop me a line.

If you really want to get into this, this weekend in Cardiff the UK-users
are meeting up for a few drinks to celebrate the 10th anniversary of FreeBSD
(which I believe is tomorrow, the 19th). I'm sure more discussion will be
on-going there. Or at least, I know me and Paul Richards will be
pontificating. :-)
 
> This frees up the size restrictions on the installer, and lets people
> without a floppy install easily, all without making anyone download
> a 600M file to do so.

The issue is, one file for anything is wrong. I *should* be able to write
the various packages (base, src, ports, manpages, etc.) as seperate files to
a CD in way I see fit and still get an installer onto a floppy that
auto-magically does the whole thing.

Think outside of the box. ISO images are *not* the way to go for install 
distributions on the long-term. Drag-and-drop however could produce a user 
ISO that would be useful for a certain class of user.

-- 
Paul Robinson