What is Hoary based on?

[HiddenWolf]

The announcement says you will be mirroring Debian 'sid' with changes where needed.

Does that mean you'll be relying mostly on Debian devels, and swapping patches around, or will you take your own path, and take from Debian the work they've already done in that direction?

IMHO the announcement sounds like 'We'll mirror Debian and add some spit, polish and artwork', while for a new distribution, I would expect it would go it's own way, after it's own goals.

Can someone elaborate?

---
* I love ubuntu, no question about it
* IMHO, it's the best compromise between Debian and Gentoo
(Options, but not the chaos/slowness of debian, and the support/friendlyness/polish of Gentoo without the compiling)
Just wanted that said, since it's not my intention to hurt anyone's feelings.

[jdong]

Ubuntu patches Debian sid packages. they're doing some weird 3-way patching to unify their changes with the Debian team's changes and the official package's changes, so Hoary will be a bit rough around the edges for a good week or two...

[HiddenWolf]

Doesn't that make Ubuntu more a debian project then a seperate distro...

Ofcourse we can't come close to Debian in numbers and infrastructure just yet, but if this would change, would Ubuntu set it's own course more?

[normnmiles]

I don't think one of the goals of Ubuntu is to provide a distinctly different distro from Debian. Instead they are focused on creating an extremely polished and somewhat up-to-date version of Debian. There is nothing wrong with this and proof of this is the popularity of other Debian based distros.

Here's a link to the offical stance on the relationship to Debian.
http://www.ubuntulinux.org/ubuntu/relationship/document_view

[daniels]

We are not a fork of Debian: it is our rock. It's quite cyclic, in that we give back to Debian regularly, and every six months, we take back from them also. Hoary is just undergoing the new stage where we drag everything back in from Debian, and attempt to re-apply the Warty customisations.

[HiddenWolf]

To me it just seems that with the state Debian is in right now, keeping up a 6 month release cycle with reasonable differences between the releases would quickly put Ubuntu far ahead of Debian, in terms of being up-to-date.

Or do I see this incorrectly?

[daniels]

To me it just seems that with the state Debian is in right now, keeping up a 6 month release cycle with reasonable differences between the releases would quickly put Ubuntu far ahead of Debian, in terms of being up-to-date.

Or do I see this incorrectly?

I'm not sure I see what you mean, but I'll try to answer it regardless. We have separate maintainence teams, so the package sets will naturally diverge -- Debian will keep picking up new versions as it does not have upstream version freezes to respect; for larger projects such as X and GNOME, Ubuntu will get newer versions first, and both will continue to apply their own tweaks to the packages. Every six months, we try to merge that, so Ubuntu gets the best of both our tweaks, and we also try to push our changes back to Debian (obviously we have a vested interest here as it makes the merge easier, so that's guaranteed to happen, altruism or not).

[HiddenWolf]

Daniels, thanks.

What I meant is that with debian being such a great work, with all those ports, and the massive amount of packages, isn't it to be expected that Ubuntu will be able to adopt packages that Debian cannot, or only months later?

The moment such a situation occurs on a major package, (say X, or gnome) wouldn't there be a risk of losing the connection to debian in that respect.

E, Ubuntu is in risk of getting such a lead on Debian, that patching is going to be hell, isn't it.

Btw, is there anyone who has a list of goals for hoary yet?
I know X.org will be in, but is it going to be monolythic or modular?
How about other refinements, anything set in stone yet?

[daniels]

What I meant is that with debian being such a great work, with all those ports, and the massive amount of packages, isn't it to be expected that Ubuntu will be able to adopt packages that Debian cannot, or only months later?

Absolutely. We had GNOME 2.8 (a first for any distribution) pretty much before it was released -- while in development, Warty always had the latest of every module. We go our own way in many respects, and then feed these improvements back to Debian.

The moment such a situation occurs on a major package, (say X, or gnome) wouldn't there be a risk of losing the connection to debian in that respect.

Already happened on GNOME, and already we are working on X.Org (I'm in Denmark right now with Fabio, our other X packager, to work on X.Org packaging).

E, Ubuntu is in risk of getting such a lead on Debian, that patching is going to be hell, isn't it.

The patching gets pretty tricky sometimes, yah.

Btw, is there anyone who has a list of goals for hoary yet?

http://wiki.ubuntu.com/HoaryHedgehog -- don't know where it is on the new wiki.

I know X.org will be in, but is it going to be monolythic or modular?

Monolithic.

How about other refinements, anything set in stone yet?

Maybe NetworkManager, definitely GNOME 2.10, probably UTF-8 related stuff, possibly a graphical installer ... we'll see.

[jdong]

Maybe NetworkManager, definitely GNOME 2.10, probably UTF-8 related stuff, possibly a graphical installer ... we'll see.

ugh, I'd rather see improvements on the current installer. It's rather effective and gets the job done; due to the low volume of questions asked, it isn't confusing.

I'd like to see the installer request some user input for X config -- i.e. even though my monitor can support 1600x1200, I prefer not to squint to read the title bars....

Lower memory requirements would be great, too -- I can't get the Ubuntu installer to "function as advertised" with 32MB RAM.

Keep in mind that even with a GUI installer, you still need a textmode fallback for older hardware, and servers with crappy video cards -- it's much more rewarding to enhance the textmode installer.

[HiddenWolf]

Monolitic X.org?

How far along is X.org with it's modular work, and will Ubuntu support that (eventually)?
Not being a devel, I could think of a few advanatges to a modular approach, how do you view this?

Are there any plans to get a graphical boot? It scared me to death the first time I booted Ubuntu and saw all the 'FATAL' hotplug errors, and for a desktop distribution, I feel the polish is important.
(I gather from the Wiki this is the case)

As to the installer, I'd rather see efficient then pretty, but for wide newbie-acceptation, it is important to go the fedora road with the installer.

As to Networking support, is it possible to increase PPP support? Perhaps add it to the installer?
There is a network here, so the installer thought there was an internet connection, because dhcp gave an IP.

[daniels]

Monolitic X.org?

Yah.

How far along is X.org with it's modular work, and will Ubuntu support that (eventually)?
Not being a devel, I could think of a few advanatges to a modular approach, how do you view this?

Personally, I am 100% behind modular X. Before I started working at Canonical, I was working part-time on a modular X server (http://debrix.freedesktop.org). We were planning to integrate modular X.Org into Hoary, giving it by far the best X packages of any distribution, ever, but unfortunately the trees just weren't ready in time -- there was a bit of parallel development with monolithic/modular trees, and I didn't have the time to merge the trees as well as package everything with Fabio. But we've both put in a lot of groundwork to make it happen, and I have packages that should be getting us modular X before Hoary's release. While they won't go in for a while, I'll be providing them as unofficial packages to test.

The monolithic structure is, frankly, total crap. It hinders our work massively, and makes it hard to maintain from an upstream and downstream point of view. The tree is bloated and out of control, and there are forks of things like FreeType and whatever in there, simply because the monolithic tree encourages the attitude of 'bugger requiring a specific version, let's just fork'. In one particular case, FreeType diverged because they had a patch to make stuff faster for CJK users which required private API (IIRC, it was a while ago). That's stupid.

Are there any plans to get a graphical boot? It scared me to death the first time I booted Ubuntu and saw all the 'FATAL' hotplug errors, and for a desktop distribution, I feel the polish is important.
(I gather from the Wiki this is the case)

Absolutely, and it will be the best boot sequence of any distro -- and *blindingly* fast! Watch Hoary for the introduction of usplash/Mad Phat Startup. :)

As to the installer, I'd rather see efficient then pretty, but for wide newbie-acceptation, it is important to go the fedora road with the installer.

We have a graphical installer planned, but in terms of usability, we feel that what we have now is actually very usable (to loosely quote Jamie Wilkinson: 'it asked me four questions, two of which were "what is your password?"'). But, that being said, a graphical installer is high on our priority list.

As to Networking support, is it possible to increase PPP support? Perhaps add it to the installer?
There is a network here, so the installer thought there was an internet connection, because dhcp gave an IP.

Ahr, hmm. I too suffered on dialup until quite recently, so I agitated for pretty good PPP support. What we have is fine in most cases, but yeah, the assumption is made that if we have an IP, we have a network, just to avoid asking questions where they really don't need to be (taking the pretty much standard case of someone who's plugged in [directly or nay] to a cable/DSL modem/router, and pops the CD in).

As a general forum question, is it possible to disable the graphical emoticons? I prefer to see a colon and a closing parenthesis, or colon P, or whatever, rather than a shiny little icon.

[daniels]

ugh, I'd rather see improvements on the current installer. It's rather effective and gets the job done; due to the low volume of questions asked, it isn't confusing.

Well, this will be based on the current installer -- imagine what you see now, but with GTKDirectFB (probably: this is not set in stone) instead of ncurses. We are absolutely not walking the anaconda route and using a GUI as an excuse to blind the user with options and shiny flashy stuff everywhere; we'll continue along the same 'simple, very few questions, totally usable' route. Jeff and I share the same metric for 'is it usable': if I gave this to my Mum, would she end up with a working Ubuntu system without having to call me? The verdict for Warty was that both his and my mum were capable of installing, and we don't want to change that.

I'd like to see the installer request some user input for X config -- i.e. even though my monitor can support 1600x1200, I prefer not to squint to read the title bars....

That's a tricky one. We try to best-guess it in most cases, but we're looking into more advanced, hand-wavy metrics (rather than just 'take the second resolution [if you have 1024x768@70 and 1024x768@60, it will use 1024x768] on CRTs and the first on LCDs', which is good to get the optimum resolution of your monitor, we're looking at stuff like checking this against the physical dimensions, mandating at least 70Hz, etc, etc) for Hoary.

What you have to bear in mind about our current system is that Fabio and I were talking about it after dinner one night at our conference in Oxford, and we started doing some vague scribbles on a whiteboard. We decided to head to bed, and started hacking on it in the morning, having thrown away all of our previous probing code. We demoed it straight after lunch.

So, we have some minor misgivings about it, but as it stands, it's actually done a surprisingly good job. The only time I've heard of it really badly falling over was when someone had a 15" monitor advertising that it was capable of 1600x1200 and 1280x1024. xresprobe picked 1280x1024, but in reality, the monitor could only do 1024x768. I offered him an old monitor of mine.

Lower memory requirements would be great, too -- I can't get the Ubuntu installer to "function as advertised" with 32MB RAM.

Hmm, what goes wrong?

Keep in mind that even with a GUI installer, you still need a textmode fallback for older hardware, and servers with crappy video cards -- it's much more rewarding to enhance the textmode installer.

Oh, absolutely. d-i has multiple frontends, and we're totally not dropping text mode support. That would be suicidal! There are also some cases where the fb gets it wrong, but they're small enough that it doesn't really matter. Our current installer uses the framebuffer anyway, so most of those bugs (very, very few) have already been beaten out.

So, I think the answer to most of your questions is that yeah, we're mad keen to do all of this stuff. Obviously we're not an infinite amount of monkeys on an infinite amount of typewriters, but judging by what we managed to do in the space of roughly three months for Warty, maybe all the stuff we've been waving away and saying 'yeah, Hoary' to for months will actually happen.

[jdong]

with 32MB RAM, the installer loops the "loading installer components" portion. Looking at the syslog reveals init repeatedly killing the process due to lack of memory. If I activate swap via console, the partitioner dies trying to umount all partitions!

[HiddenWolf]

Daniels:

I totally agree with the 'ask as little questions as possbile' approach.
However:
While not asking the questions is ok, I feel for something as essential as network, the installer should confirm a connection, IE ping www.ubuntu.com, whatever.
If this does not succeed, the user should be warned, and given options to configure/solve this, before the installer enters an endless round of timing out on apt-get.

As for fast, make it priority number one :-)
Modular X would be great, since then it looks to be far easier to overhaul parts of the system witthout the infamous big bang. :-)

[hasan]

Absolutely. We had GNOME 2.8 (a first for any distribution) pretty much before it was released -- while in development, Warty always had the latest of every module. We go our own way in many respects, and then feed these improvements back to Debian.

IMHO the only reason ubuntu released or let me put it this way, the only reason Ubuntu had gnome 2.8 was because it was a planned marketing strategy. Ubuntu developers knew that if Ubuntu was the first distro to have gnome 2.8 (people know how diffucult it is to install gnome) it will atract alot of attention from the gnome users ultimatley making flash news on slashdot or simillar.

I like ubuntu... Hoary is great but.. i think its time Ubuntu split its branches with debian and do something useful with themselves.

(these are opinions not based on monkey logic, no animals were hurt in the process neither)

[daniels]

IMHO the only reason ubuntu released or let me put it this way, the only reason Ubuntu had gnome 2.8 was because it was a planned marketing strategy. Ubuntu developers knew that if Ubuntu was the first distro to have gnome 2.8 (people know how diffucult it is to install gnome) it will atract alot of attention from the gnome users ultimatley making flash news on slashdot or simillar.

I like ubuntu... Hoary is great but.. i think its time Ubuntu split its branches with debian and do something useful with themselves.Much of our strength comes from Debian, and we are reliant on it. Say what you will about Debian, but it has an amazingly solid base from which to work from, great support for everything, a fantastic toolchain, and absolutely *everything* is packaged.

(Disclaimer: I am also a Debian developer, albeit a rather latent one of late.)

[Lovechild]

I do hope that the new bootsplash won't print out alot of information like the rhgb in FC3 does, it's completely useless for 99% of all users, it should just be a progress bar that, preferredly, would fly past directly from grub into gdm, no text printed, as even booting with quiet gets us a little text we should hack that out as well in a nice way of course, no unneeded information displayed (like printing subsystem started, who cares, if it doesn't start it's a bug so don't bother me about it while I boot - KISS)

Otherwise it's good to see UTF-8 being the new standard, that will make us translators happy, by the way, where are the ubuntu specific files for translation located, I would like to get to work on them as soon as possible to get Hoary goodness in 100% danish.

I also hope to see some kind of system to give me danish out of the box firefox and openoffice, it's slightly annoying to have to set it manually.

And a rhn-applet like updater would be neat, just sitting in the notifcation area to alert me about updates, or using the proposed message notification system for GNOME (like that's going to be ready for Hoary.. *sad*)