[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Re: slink is gone, goals for potato?


On Tue, 2 Mar 1999, Edward Betts wrote:

> So slink is released, it is stable, everybody is happy. Kids are
> dancing in the streets, and we are all a little out of it on
> {v,d}beer. You get the picture. 

Well, I'm busy going through panic maneuvers. Got a call at 6p tonight.
"We need you in tomorrow. Forget two weeks. Tomorrow." "Uhm, okay. Pay
starts then too?" "Yes." "See ya tomorrow." ;P
> But what about potato? No, I am not talking about the stuff they make vodka
> out of, I am talking about the next release. Here is some ideas I throw
> together, but I think our main goal should be to have no release goals :-)

Well, that'd be a beautiful thing. But I'm not big on chaos. ;)
> General project goals:
> potato estimated release - between June and August 1999

I think that's rushing a bit. I'm more inclined to rule out June unless
things go absolutely spectacular. ;)

> EGCS as default C compiler on all architectures
>   It is needed to compile glibc 2.1

...yet as I understand it, 2.2.x doesn't quite like it. Nor do many other
programs. I'm gonna cast my vote to do egcs optional, with gcc default. We
can build glibc 2.1 with egcs, and still use gcc.

> PAM as standard for authenticifcation.

I've looked at PAM. I'm all for it. Definitely.

> Filesystem Hierarchy Standard (FHS)
>   Changes needed, off the top of my head - 
>     * move /usr/doc/*    to /usr/share/doc/*
>     * move /usr/info/*   to /usr/share/info/*
>     * move /usr/man/*    to /usr/share/man/*
>     * move /var/lib/dpkg to /var/state/dpkg
>     * move /var/lib/*    to /var/?????/*
>     * move non-lib dirs out of /usr/lib/* to /usr/share/*
>     * remove /usr/dict
>   How to hander this? 
>     * Set lintain to complian
>     * Just start uploading with changed dirs? and hope every package makes an 
>       upload before release
>     * Start filing bugs for packages that have not been changed
>     * Do some NMUs
>   What to do with old dirs? Leave symlinks to new ones or not?

Okay, I'm big on FS organization. But I like things in sane places.

	* /usr/doc	to /usr/share/doc
	* /usr/info	to /usr/share/info
	* /usr/man	to /usr/share/man
	* /var/lib/dpkg	to /var/state/dpkg
	* /var/apt	to /usr/apt

apt keeps it's cache in /var. I make small /var partitions, usually <500M.
I keep /var small. Most people I know do. Doing an apt-get dist-upgrade
with a tiny /var partition is NOT easy. I had to symlink. So move it. ;)

I'd say let's keep the old dirs, as symlinks, too. It takes time to adjust
to new locations. Let's keep the libs where they are. Non-lib stuff in
/usr/share is okay by me I guess.

> First public release for powerpc architecture

Assuming I can get the machines soon, I may be able to get started on at
least some PowerPC work. Like I said; I start the new job tomorrow. ;)

> Completely support package uploads signed with gpg?

Yes. Yes.

> Support for lsh instead of ssh (lsh needs some work)

Support for lsh *and* ssh. Ssh has Windows and DOS clients. I see no
reason to abandon it.

> Package management goals:
> Apt, the new front end to the package management system as standard
>   Versions for:
>     * X, kde/qt      - someone will do a port

kde/qt is non-free. Why oh *why* should we make an apt frontend for
kde/qt? They can deal with the standard one that uses generic libs. The
only reason I'm still for a gnome-specific one is because gnome is free,
and it has some really neat features.

> Install from serial line

Sounds good. apt install would be nice too, if it's not in slink. I
haven't checked. I upgraded hamm->slink.

> More languages for the boot disk 

Already determined to be nigh-impossible. Perhaps language specific
bootdisks? Definitely more languages on the CD, since we can write it
bootable. ;)

> Graphical install Ben Pfaff is working on (optional)

IIRC, RedHat 4.2 had an X-based graphical install. Perhaps we should take
a look at how they did that, and follow their example?

>   Kernel boot logo? - Debian logo

Get me the logo, I can cover that. :)  I collect obscure kernel patches, I
have one for 2.2.x and one for 2.0.x that'll do it. :)

> CD-ROMs with 2.88Mb boot disk, fit more stuff in.
> Start work on boot disks and CD early, so ready for release, no mad rush.

Gonna have a mad-rush anyways, probably. ;)
> Programs to include:
> Kernel
>   Linux kernel 2.2.x  - default
>   Linux 2.0.3[3-7]    - an option for those who do not want to run 2.2


>   Hurd                - possibly?

I'd cast that as a 'most likely'

>   {Free,Open,Net}BSD  - unlikely

We need to change this, folks. Whether or not we have the support of the
world behind us, so long as it's okay by whichever BSD is used, that's one
more accomplishment we can show off.

> Libraries
>   glibc 2.1           - lots of recompiles

I won't support glibc 2.1 fully till licensing issues are resolved. And
GNU won't release it till those issues are resolved. So that's up in the
air for now at least.

>   glib and gtk+ 1.2   - lots of recompiles
> Desktop enviroments
>   Gnome 1.x           - Default?
>   KDE 1.1 (free QT)   - No hard feelings?

I'm sorry, but QT is not GPL, therefore at *best* it gets to go in
contrib, more likely non-free, as it is now. TrollTech will never GPL QT.

>   GNUstep             - Will it be stable?
>   UDE                 - Is it free?

Somebody needs to look into UDE. 

> Window Managers
>   Enlightenment DR 15 - still developers release, not 1.0, but stable

A definite 'maybe.' Personally, I love E. The only question is, are they
going to have a stable tree (non-snapshot) in time?

> Graphical Users Interfaces
>   xfree86 3.3.3.x     - some more drivers, with redhat xconfigurator?

xf86config, XF86Setup, xf98config, XF98Setup. That's all we really need. I
see absolutely NO reason to use RedHat's Xconfigurator. Furthermore, I
think that's MetroX specific, but I'm not about to build a RedHat system
just to confirm it either way. 

>   berlin 0.x          - just for those who want to play

I'd vote to stick it in 'contrib,' for that reason. 

> perl 5.005            - lots of recompiles
> GNU Emacs 20.5        - new display engine with console syntax colouring
> Mozilla 5             - it should be released in time

I wouldn't bet on it.

> inn 2                 - maybe the old version avaliable as well

Witnessed inn 2.x breakage. Going to have to keep 1.x available as well.

> Have I missed any other goals or important points that need discussing?

Well, first things first. EVERYONE knows my bias by now. Besides first
release of Debian GNU/Linux-PowerPC, I'd like to see boot disks for
RS/6000's, as well as install documentation, as well as possibly a script
for creating bootable restore tapes for RS/6000's that support it. (IBM
DDS, DDS2, DDS3 tape drives support bootable tapes. Trivial to do it, I
think.:) I'll be more than glad to do what I can to the documentation of
install, as well as the bootdisk ends. But I need documentation and
testing platforms for bootable restore tapes. It MAY require some work to
mt/tar, perhaps a break-off 'bmt/btar' set, b for bootable and all. ;)

Next up, maybe it's just something I passed, but I didn't see a package
with the many licenses of Debian. I think we should have a base package,
'debian-licenses,' of REQUIRED priority, that simply contains all the
Debian-specific licenses, and other well-known licenses, such as the GPL,
BSD, etcetera. 

Another thing; I'm not sure if pppconfig is maintained upstream. I haven't
had a chance to look. But I've submitting a 'wishlist' bug against it. It
doesn't have the ability to "manage" connections. Ie; delete, modify,
etcetera. That has to be done by hand. I'd like to see at least 'modify'
abilities in it for potato. 

> I can't wait until my slink CDs come in the post, I suppose they still
> have to press them first.

*laugh* I have to burn them. HOPEFULLY I can snag a CD-R or two tomorrow,
and get those things burned in a reasonable amount of time. ;)

| Phillip R. Jaenke       | "Not all wisdom comes from without;  |
| prj@nls.net             |  much wisdom can only come from      |
| InterNIC: PRJ5          |  within. Only you can teach yourself |
| Professional Unix Guru  |  some of the most important lessons  | 
+-------------------------+  of life." --Takes-Many-Roads,       |
Project Head              |              Silent Strider Theurge  |
the Linux-RS/6000 Project +--------------------------------------+
- - http://www.nls.net/mp/prj/linux/ -- http://www.nls.net/mp/prj/ -

Version: 2.6.3a
Charset: noconv


Reply to: