Slashdot Mirror
Mark Shuttleworth Proposes Delaying next Ubuntu
Beuno writes "Mark Shuttleworth has proposed on the ubuntu-art mailing list to postpone the 'Dapper Drake' release by 6 weeks. He lays out the reasons pretty clearly: the delay should make the release a more user-friendly distro. He has also called up a community meeting in April 14th on IRC for community input. Is it really worth delaying the release for more then a month just to polish it out a little bit?" Commentary on this also available from the Tectonic site.
He proposes a town hall for March 14, not April.
How long exactly has Longhorn, er, Vista been pushed off? Six weeks pales in comparison.
Is it really worth delaying the release for more then a month just to polish it out a little bit?
Absolutely.
505 users in favor of the delay, 50 against at last count.
http://ubuntuforums.org/showthread.php?t=142536
Dapper is coming along nicely, but there are a number of bugs that might not get the attention they deserve if Dapper is released on schedule.
Their Flight 5 CD is out. It should be quite stable for normal use.
As a Gentoo user, I tried out Ubuntu on an old Toshiba laptop about a month or three ago when the current version came out. I liked what a I saw, but I ran into to huge problems. One, Ubuntu completely screwed up the monitor settings for the laptop, and the sound was completely futzed. I found the solution to fixing the monitor settings on an Ubuntu user forum (involved hand editing X.org's conf file) and the sound, well, I managed to get it to play somewhat but GNOME still never detected it properly.
If Ubuntu wants to be "Linux for human beings" it needs all the polish it can get after that experience.
Keep up the good work guys.
A 6 week delay doesn't sound earth shattering to me... I fail to see the problem here, to be honest. Especially if it's about improving usability, an area critical for Linux adoption, which is one of the main purposes for this particular distro.
To me, this feels basically like delaying an extra security heavy distro 6 weeks to implement verify a new security protocol implementation works correctly.
Beware: In C++, your friends can see your privates!
Considering that they want this to be the first Ubuntu release that's supported for a long time and that can compete with things like SuSE's or RedHat's enterprise distributions, I'd say six weeks are perfectly acceptable.
quidquid latine dictum sit altum videtur.
To be honest, Dapper is very stable and polished already. There's mixed reactions over the new Clearlooks scheme they've implemented but overall, it's turning out very well. I can't speak for the localisation issues, but a stable release is much better than a rushed release. If you want to try Dapper, Flight 5 should be just fine.
If we can hit that bull's-eye, the rest of the dominoes will fall like a house of cards... Checkmate.
>Is it really worth delaying the release for more then a month just to polish it out a little bit?
there are hundreds of distros already, and the only thing they all lack is polish, so yes.
what's the hurry?
Patience is a virtue. Ubuntu has no need to generate revenue, and if it takes six more weeks to make the release more usable for human beings, that can only be a good thing.
If it wasn't for the fact that Ubuntu is synchronized with Gnome releases I wouldn't mind the delay. But now they would have to either rush the next release, be late with it or completly skip Gnome 2.16. I hope they'll find some good solution because many users are preferring Ubuntu to other distros because of fairly nice bleeding-edgeness. With this step they could lose major selling point to causal Linux geeks.
The recent theme changes are not a step in good direction too. It looks abysymal and burns my eyes. Even tough I didn't like brown theme the new one made me miss it.
I think that almost everybody would agree that a little more time spent making a product better is a good thing.
:) Any steps, no matter how small, to appeal to the Chinese/Korean/Japanese markets will probably pay off well.
:)
It's not just about polish, though. TFA lays out a number of points where improvements are needed:
1. Testing
2. Certification
3. Localisation
4. (last but not least) Polish
Improvements to Asian localisation should help a ton of people - we're not all English speakers.
Not that it all matters to me, though... I use SUSE.
BaltikaTroika
The fact that you're not a software engineer shows.
Want to know what would have otherwise loaded? The Windows Bootloader, which would have been within the exact same 512b sector that Grub now occupies. Boot loaders on PCs are extremely restricted in what they can do -- their code can be no larger than 446b in size, they run in real mode, and basically must rely directly on BIOS for all of their I/O routines.
In effect, this is 1980's technology, and flexability is virtually nil. The primary boot loader can't just pass its duties off to another boot loader, as there aren't really sufficient instructions available to do this, and the two boot loaders cannot occupy the same space on the drive.
If you're looking for something to blame for this situation, it's the fact that the architecture of the PC BIOS hasn't changed significantly in more than 20 years. It's still firmly rooted in the days of 160KB floppy booting, where the idea of a second-stage boot loader for choosing what OS you want to boot would never have occurred (want to boot a different OS on a diskette-only system? Use a different boot disk). BIOS should have died a long time ago.
Boot loaders like GRUB do the best they can with what little resources and possibilities they are given. I'm sorry that the GRUB developers don't have access to your screwy system to test and debug on. Here I've run GRUB on a variety of systems, and the only machine I ever found which had problems with it is one with a built-in nVidia chipset, back in the Fedora Core 2 days, which was easily solved by switching to a different boot loader.
Yaz.
What does Ubuntu offer that Debian doesn't?
a reasonablly predictable release schedule (a bit too fast for my liking in fact) and a bit of polish for some desktop related stuff.
as such it fills the gap between debian stable (slow unpredictable release process) and debian testing (constant upgrade treadmill with little in the way of security support)
What can be done with Ubuntu that I can't do with Debian?
if you feel like supporting debian testing/unstable then nothing. And with sarge for a while probablly not much.
However in the couple of years prior to the sarge release running woody was becoming more and more untenable as recent software simply wasn't getting tested with stuff that old. Sarge is ok for the moment but unless debian can get thier house in order and come up with a release every few years at least then we are going to run into this issue again.
note: i'm known as plugwash most places but i screwd up registering that here somehow in the past and now can't register
Yeah, thats a common misconception. Ubuntu is not a snapshot of Debian Unstable. Multiverse is a snapshot of packages in Debian Unstable that are not in Universe, Main, or Restricted. Universe contains packages supported by the community, which is encouraged to work closely with Debian. Main and Restricted are both modules that are directly supported by Canonical. These packages are worked on heavily by employees of Canonical and while there is significant collaboration(some would like to see more, but thats a seperate debate) these packages are not just stabalized snapshots. Canonical puts a lot of time into Main and Restricted and you will often see versions of packages(and packages) that are in Ubuntu before they hit Debian. You can see that by the fact that Ubuntu Dapper currently has the prerelease gnome 2.13 while Debian still has 2.12. Please stop spreading this misinformation.
"We Don't Need No Truthless Heros!" - Project 86
By the way: does anyone know if dapper will ship a kernel that's been compiled with the version of gcc that's included on the distribution CD's? If badger had one fundamental flaw, it would be a kernel compiled with gcc3.4 and gcc 4.? included on the CD. People who need to compile e.g. their wireless driver because it's not included in the standard kernel, are fucked, because they may not have network access with the distribution files and need to download either gcc 3.4 or kernel sources...
As many others have pointed out, in 446 bytes, we can't do anything. All the Microsoft boot loader have historically done when it barfs is print something like "NT Loader not Found", and then left you "locked out of your system", just as GRUB did.
BTW, you're not really locked out. You can create a GRUB boot floppy and manually boot into your OS installation. You can also use the Windows CD to set the MBR back to its original state. Or you could use most Linux distros' rescue CDs to fix the problem.
Well, more naked people for one.
> spits control back to whatever would load in the absence of GRUB having been installed.
/boot/grub/stage2. GRUB needs to do this because it is bigger than 512 bytes, so stage 2 contains all of the GRUB code that doesn't fit in the first 512. GRUB needs to be this larger than 512 bytes because it's a really advanced boot loader, it even understands file systems, which allows it to load configuration files, initrds, kernels, and modules by reading the file system, instead of having hard coded locations of those files location (by disk geometry) rammed into it. (this really helps when you update, replace, or change those files!)
/boot/grub/stage2, a location which is hard coded with the disk geometry location of this file. (stage 1 doesn't understand file systems).
/boot/grub/stage2, the parition table, and the boot sector of your windows partition are completely differnt locations on the drive, it is entirely posible that GRUB stage 1 could have a problem, while the Microsoft MBR could not.
/boot/grub/stage2 is located, but not in the partition table or boot sector of the Windows partition, one could fail where the other succeds.
/boot/grub/stage2) is loadable, but a bug in the implementation means it only works for block_number=(location of partition table) or block_number=(location of boot sector). Who wants to bet that there are BIOS out there that only get tested by the manufacturer on MBRs that only load play with the partition table and boot sectors of partiti
The BIOS knows you want to boot from your hard drive, it does one simple thing to facilitate this, it loads the first 512 bytes from the drive into memory, and it tells the CPU "start executing here". Should the code in those 512 bytes fail, the bios has nothing further it can do, it only knows how to do one thing, grab the 512 bytes and let them execute.
You installed Stage 1 of GRUB in the MBR (first 512 bytes of the drive). When you installed it, you installed it over top of the 512 bytes that were Microsoft's MBR. This is what was there before GRUB was installed, and now it is gone, completely written over, and neither GRUB nor the bios can do anything about it.
I think you would probably like it if the grub installer put a backup copy of the Microsoft MBR somewhere else on the drive, and you would like stage 1 of GRUB to load and execute those if there is any problem. But, if there is an error loading those 512 bytes, absolutely nothing can be done.
There is a perfectly valid explanation for why stage 1 might fail and why the microsoft MBR doesn't.
Stage 1 of GRUB (installed in the mbr) has 1 job, load a file from your Ubuntu partition,
The Microsoft MBR also has a simple job. It looks, at the partition table for partitions marked as bootable, takes the first one, loads the boot sector of that partition into memory, and executes it.
So stage 1 of GRUB and the Microsoft MBR really have a lot in common, as they are both 512 bytes they really do shit all, they just attempt to load more boot code off the drive and let it rip. The crucial difference here is WHERE on the drive they play with. Microsoft MBR reads the partition table and the boot sector of the partition marked bootable. GRUB stage 1 reads the location of
As
What could be different about these different locations on the drive?
If there was an error on the drive where
Or, maybe the hard drive is fine in all locations, but the mechanism used by these two MBRs to access it is not behaving as it should. What is this mechanism? Our frequenly buggy friend, the BIOS. The BIOS implements a interface that the MBR can use to get its job done. Something like
load_sector_from_ide_drive( ide_channel, master_or_slave, block_number )
Assume neither MBR has any bugs in calling this interface, what if there is a problem with the implementation itself? What if the interface promises that a block_number=(location of
I'm Shocked, shocked I tell you that a distro based on, and dependent upon, Debian packages would choose to focus on some kind of abstract "usability" or "stability" issue over fast and frequent updates!
Where's the bleeding edge code? Where's the "It compiled this morning let's push it out" mentality that's so common with Debian based Distros??
I'm astounded and saddened. Microsoft has updates coming out weekly. It can't be good for Ubuntu if it loses the "update war" with Microsoft. If you lose the update war, everything else is down hill from there.
"Live Free or Die." Don't like it? Then keep out of the USA
I remember keeping track of the Breezy Badger planning wiki before that version was released, and it seemed to me that the team deferred many of their major goals... on the other hand, it looks like most improvements planned for Dapper have been implemented already, as Shuttleworth notes in his message:
s
https://launchpad.net/distros/ubuntu/dapper/+spec
I'll refrain from Debian comparisons, as they're not needed to communicate what stellar work the team has done here. Point is, Ubuntu users and admins ought to support this delay, for the same reason I support Ubuntu... the Ubuntu team simply has its shit together, moreso than that of any other freely available distribution.
Let Shuttleworth strategize to take on Red Hat, SuSE, and Vista--because Ubuntu actually has a fighting chance. That prospect ought to excite Ubuntu partisans (like me) and fence-sitters alike.
Amen to that! I tried installing Ubuntu on my girlfriend's laptop, and in the end I just gave up getting Chinese input working properly (she's Taiwanese and sends a lot of mail in Chinese to her friends back home.) After a couple of long nights spent fiddling with it, I could get it to sort of work with some apps, but this is one area where Windows beats Linux hands down -- after I gave up and installed Windows on her machine, enabling Chinese input took me all of about 30 seconds to do, and it works flawlessly in every app she uses.
I used to use Debian on my laptop, but later switched to Ubuntu. Why?
It supports more hardware out-of-the-box, and it has newer GNOME packages than Debian. Things that I had to install in Debian (the touchpad, etc) were already installed.
I wouldn't use Ubuntu on a server though, everything I can do in Ubuntu I can do in Debian. Installing a Debian desktop is just more hassle than installing Ubuntu.