[indiana-discuss] Tim's 0.02

Dirk Wetter dirk.wetter at drwetter.org
Sun Jun 10 05:51:32 PDT 2007


Hi there,

Am 06/10/2007 01:00 PM, Tim Foster schrieb:
> Hi All,
> 
> I'm really looking forward to seeing that Indiana can offer, and will
> do anything I can to help out.
> 
> My thoughts:
> 
>  - Package management
> 
>  I echo the comments so far: a network-based package repository,
> that's smart enough not to re-install libraries already on my system
> would be super. 1 CD to install would be great too.

+1. I like the way Ubuntu does this. DVD would be also fine, considering
people who don't have as much bandwidth as people in most parts of the world.

*One* mechanism how to get packages installed on the system is needed, as
e.g. Nexenta is doing it. Not a combination of blastwave, sunfreeware and
others.

Also it should contain package classes determined by the usage of the system
(e.g. desktop, laptop, LDAP server, DNS/DHCP server, development system, ..)

You can consider dropping ancient CDE. Maybe support KDE is a good idea, too.


>  - Install
> 
>  I know there's work ongoing on a new installer - but I was really
> excited by the comments Ian made about virtualisation during his SVOSUG
> talk[1]. I suspect it'll be a while before we get this, but it sounds
> fantastic.

Haven't seen the new installer yet. I am coming from the Linux and Solaris/Sparc
world (for 11+ years and been writing tech reports since 12+ years for Linux
distros). Doing my first steps on OpenSolaris/x86 is somewhat a pain. It is
certainly more real for beginners, if you have more than one system already
installed. This is important if the user base should not be geeks only:

A) Osol/x86 is just too picky where it wants to get installed.
B) The bootloader not really cooperative yet, too.
C) An installer should give an overview what's going to happen next and what
steps already were done (Suse and Xandros do a good task with respect to this).

>  - Desktop
> 
>  It might be worthwhile for everyone to read the Desktop Gaps document
> that the Desktop Community put together a while back:
> 
> http://www.opensolaris.org/os/project/jds/documents/desktop_gaps
> 
> I think it pretty clearly shows the areas that are lacking in the
> OpenSolaris desktop today.

Good overview!

> Of course, many are already being looked at as part of the ongoing JDS
> work, which is great to see - the desktop has come on in leaps & bounds
> over the last few years - it's *nearly* at the stage where I'd be happy
> to ditch my Mac at home in favour of OpenSolaris.

If you want to broaden the user base, desktop is certainly important. However
it's difficult since the variety of hardware is also bigger as opposed to
servers. OpenSolaris -- sorry for my clear words, but it doesn't help the
matter otherwise -- is concerning solid standard hardware (e.g. from Intel) at
least half a year behind Linux regarding supported hardware and drivers for
that. There is not even a bluetooth stack and driver support in the kernel.

I know that for Linux, the more you go to the cheapish hardware, the air is
getting thinner. You should ask yourself now how many people are buying this
sort of hardware and how many people will be annoyed by trying OpenSolaris and
then figuring out that there's no support for their beloved NIC/hard disk
controller/video card/...

Bottom line: Desktop hardware is really really difficult to support unless you
have a lot of man power. BUt certainly desktop is what draws people.

>  - Power management
> 
>  For me, getting suspend-to-ram working would be enough to make that
> switch. Solaris already archives all my photos, music, etc. but being
> able to get instant-on would promote Solaris to being my main platform
> for nearly all my home computer use.

Well, yes. For some people laptop usage is more or less useless if you can't
suspend the beast ;-)

ACPI suspend, specifically suspend to RAM was a pain for a long time under Linux
kernel 2.6. Before that there was no ACPI IIRC (only backports from e.g. Suse),
only APM. I am talking about years here, and there were a lot of people
dedicated working on it. And it still has problems under certain, mostly older
hardware. Also CPU throtteling took some time to get the most out of the
batteries.

As I said, excuse me if I wasn't very polite with what I wrote. But I think the
kid-glove kind of apporach doesn't help.


My 0.02 €,

	Dirk


Dr. Wetter IT Consulting                         http://drwetter.org
IT Security + Open Source
Key fingerprint = 80A2 742B 8195 969C 5FA6  6584 8B6E 59C1 E41B 9153








More information about the indiana-discuss mailing list