On Thu, 26 Oct 2000, Derek Martin wrote:
> On Thu, 26 Oct 2000, Rich Payne wrote:
>
> > > > Interesting....I always thought glibc was a significant part of the
> > > > distribution....seeing as almost every program makes use of it....
> > >
> > > I _did_ say not MUCH significant, not NOTHING significant. And honestly
> > > Rich, how many people even notice that glibc has been upgraded? Plus,
> > > didn't I hear that it too was buggy?
> >
> > How can say that glibc isn't significant when just about every package
> > present on the CD(s) use it? As for how many people, it depends on what
> > you are interested in. As for other things updated:
>
> Rich, I'm not saying that at all... but on the other hand, what about
> glibc has changed significantly as it affects the average Linux user?
> IOW, why, as an average Linux user, would I want to upgrade my glibc from
> 2.1.3-19 (the RH 6.2 bug fix) to whatever is offered by RH7? And what
> version is that, BTW?
Yes, but updates to glibc are always small and incremental....you can't go
making massive changes to something like that (at least no easily). So by
your logic there would never be a reason for a new distribution, because
the changes are usually small (unless of course it fixes a bug that
effects you, in which it's probably a big deal).
>
> > GNOME (1.2 vs 1.0)
>
> Hmm... It's been a while since I actually used 1.0 since I've found Helix,
> but were the changes really all that significant? Mostly some bug fixes,
> weren't they? Even if they were, that wasn't really the gist of my point
> (which is coming, eventually)...
>
>
> > adding openssh/openssl
> > new installer that can use multi-CDs, which means more packages available
> > how about a newer kernel with USB, AGPGart support
> > much newer PCMCIA code
> > added MySQL, xemacs, new gimp....plus testing version of KDE, KOffice,
> > Glide....Mesa
> > RPM V4 (not sure of the benefit there though....)
>
> In any case, how significant these are is largely subjective. For
> example, I would consider the kernel changes significant, but that's
> hardly worth buying a whole new distro for. The PCMCIA stuff isn't part
> of the kernel proper, is very small, and can be upgraded rather quickly
> and painlessly if you know how to build a kernel, so it isn't too
> significant to me (though it might be to some, especially new users with
> PCMCIA stuff that's not supported by the older code, but you've already
> said you don't recommend RH7 for newbies, so that doesn't support your
> argument at all).
Again, based on your argument there would never be a reason for a new
version. After all, most of thie stuff is available seperatly, so I can
just take my RedHat 6.2 box and keep appling updates/new software.
> What I was attempting to point out is that RH is under pressure to sell
> distros, so they need to appeal to 1) new users, who hopefully (from
> RedHat's perspective) will buy their stuff as their first distro, and 2)
> people who already have previous RedHat distros, who have been happy with
> them, and want to get the latest stuff with the least amount of hassle.
> They want to have the "latest and greatest" of what's out there to appeal
> to both. But as it happens, right now a lot of that stuff is beta (or
> even alpha) quality. That's what my original message was meant to
> address. I don't blame RedHat for it so much... they are after all a
> commercial company trying to thrive on software that's essentially free,
> and the way they do that is by selling distros (and support too).
Come on...everything in Linux in beta.....nothing's ever 'finished'.
As for selling CDs...I'd be curious to how much of their income comes from
box sets. My guess would be fairly small...the moneys in the support.
> In general, I think the differences between their point releases are akin
> to the differences between W95 and W98; i.e. largely cosmetic with some
> bug fixes, but with lots of other broken stuff. I expect more than that
> from a Linux distro... both from my perspective as an experienced Linux
> user, who relies on Linux to be stable, high-quality software, and from
> the perspective of attracting new users to Linux, for much the same
> reasons.
and people payed $99 for the Win98 update..7.0 is a bargain! (Sorry,
couldn't resist).
> Whether you or I would recommend 7.0 to newbies, they are a large part of
> the group that RH is trying to attract, which I think in this case would
> be unfortunate. Remember when you were a newbie? Did you have a lot of
> friends that already used Linux, who could offer you intelligent
> information about how to chose a distro? I didn't... and neither did a
> lot of people in the early goings. And a lot of present-day new users
> won't either. But if they've done any research, they've probably heard
> that RH is a great distro for beginners. I also agree that Mandrake is
> better for newbies, and their increase in market share suggests that we're
> not alone. But RH still ranks very highly in most such polls, and if you
> ask me a beginner faced with RH7 runs the risk of thinking that Linux
> isn't any better than windows in terms of reliability and quality, and I
> for one don't want to see that happen.
>
> > RH shipped 3.3.6 servers for cards that don't work well under 4.x. Some I
> > think you can even manually pick which you would like.
>
> I'm glad to hear that. At least they're thinking about those issues...
>
>
> > > I think their best option would have been to ship both XF4 and XF3
> > > and give the user a choice, with a warning that XF4 may not work for them,
> > > especially if they have older cards.
> >
> > I think they looked into this....but there was some fundamental
> > differences that made this not possible. Mainly to do with the Xlibraries
> > that various apps were compiled against.
>
> If the libs for XF4 aren't backward compatible, then I guess that's a good
> point. But then I upgraded from XF3.3.6 to XF4.0.1 without any hassle, so
> I have some trouble buying that...
>
>
> > I you don't like using newer code then go back to Linux 2.0 and
> > libc5. After all they were stable. People bitched like there was no
> > tomorrow when RH moved to glibc. Yes, it hurt, but guess what. All the
> > archs are now on the same version of Glibc (give or take minor revs) and
> > things are much simpler. If RH had not done that, who would have? Would
> > i386 Linux still be using libc5 when Alpha and Sparc were on Glibc2?
>
> >
> > I think they looked into this....but there was some fundamental
> > differences that made this not possible. Mainly to do with the Xlibraries
> > that various apps were compiled against.
>
> If the libs for XF4 aren't backward compatible, then I guess that's a good
> point. But then I upgraded from XF3.3.6 to XF4.0.1 without any hassle, so
> I have some trouble buying that...
Yes, same here but they did have a good reason, wish I'd saved that email
now.
>
> > I you don't like using newer code then go back to Linux 2.0 and
> > libc5. After all they were stable. People bitched like there was no
> > tomorrow when RH moved to glibc. Yes, it hurt, but guess what. All the
> > archs are now on the same version of Glibc (give or take minor revs) and
> > things are much simpler. If RH had not done that, who would have? Would
> > i386 Linux still be using libc5 when Alpha and Sparc were on Glibc2?
>
> I agree that eventually you have to move on. But see my earlier comments
> about relying on stability etc... My hope is that we avoid the reputation
> that Windows has, but I fear that we won't because of the very same
> pressures that Microsoft experiences, exerting pressure on the commercial
> distros to release new software.
What would be interesting here would be to look at RH's release cycle for
the past several years and see if they are really speeding it up. I don't
think they are, but I don't have the release dates handy.
>
> I think it depends a lot. I still can do a complete RH install in about
> 10 minutes, provided I have a system which has only hardware which is well
> supported. Granted, I can do that because I have enough experience with
> Linux that I can chose such components to put in my systems, but
> nevertheless I don't think there is any other distro that I could install
> that fast.
>
> But, a fast install time isn't the most important factor in chosing a
> distro...
Well, it could be....Kickstart is a powerful thing (when it works). We've
isntalled clusters of Alphas via NFS root and kickstart. I think it was
about 3-4 minutes per machine for the install :)
--rdp
--
Rich Payne
[EMAIL PROTECTED] www.alphalinux.org
**********************************************************
To unsubscribe from this list, send mail to
[EMAIL PROTECTED] with the following text in the
*body* (*not* the subject line) of the letter:
unsubscribe gnhlug
**********************************************************
