- Site menu:
- Latest Discussions
- Latest Newswire
- Archives
- Submit Story
- Story HOWTO
- Syndicate
- E-mail Alerts
How Canonical Can Do Ubuntu Right: It Isn't a Technical Problem
|
|
|
Comments (37)
17 earlier comments in the full thread.
The other problem with Firefox is plugins that get broken with Firefox updates. You have your fave extensions, use them a lot, then Firefox issues a "you must update or the sky will fall!" announcement. After the update your fave extensions don't work anymore. Makes me wonder what kind of craziness is going on under the hood.
>... it's pretty obvious that Ubuntu's QA is horrible if people have to do this.
Horrible is a rather strong word. Ubuntu's QA compares quite favorably with most closed source software I've used, for example. I reserve horrible for those software companies which actually deserve it, say Adobe and Microsoft.
> The fact that a majority of users out there _DON'T_ use something else means they are either ultimately stupid or a glutton for punishment.
Or they're former Windows users, in which case they're used to it.
> I'm having just such a problem with Firefox 3.6.x - I have a Web-based app that works with 3.5.x but not 3.6.x.
While my online banking site works fine with 3.0 and 3.6, but won't work at all with 3.5. :(
Horrible is a rather strong word. Ubuntu's QA compares quite favorably with most closed source software I've used, for example. I reserve horrible for those software companies which actually deserve it, say Adobe and Microsoft.
> The fact that a majority of users out there _DON'T_ use something else means they are either ultimately stupid or a glutton for punishment.
Or they're former Windows users, in which case they're used to it.
> I'm having just such a problem with Firefox 3.6.x - I have a Web-based app that works with 3.5.x but not 3.6.x.
While my online banking site works fine with 3.0 and 3.6, but won't work at all with 3.5. :(
Quoting:2. Everyone (mostly) thinks not fixing these bugs is fine (or not Ubuntu's fault)
I don't think this is fine actually. It's fine for Debian because they "release when ready". It's not fine for Ubuntu with a 6 month deadline.
Quoting:The thing with Debian, though, is they're releasing when "ready." But what constitutes ready?
If everything on the install image - say in the default desktop - meets this standard, than that is very powerful indeed.
It's more powerful. It applies to the whole 25,000+ package repository. All of that is Debian. It's not split like in the Ubuntu world where you have supported and contributed applications. Debian releases all of it, or none of it. What happens to be on the CD is an entirely different matter. Especially if you consider that many people use the net-install CD which only carries the minimum required packages needed to download everything from the net.
Quoting:I don't know what sysadmins are doing these days, but way back if they wanted to run Apache, they compiled it themselves with exactly the version and features they wanted. Maybe today there is greater use of precompiled packages even on servers. It sure makes installation and patching easier.
I think that many will use precompiled packages, or they compile their own and stuff them in a separate repository. Luckily software like Apache has become much more modular. You can simply load and unload modules at runtime. You don't even need to restart! So there is a lower need for compiling your own.
Quoting:The other problem with Firefox is plugins that get broken with Firefox updates. You have your fave extensions, use them a lot, then Firefox issues a "you must update or the sky will fall!" announcement. After the update your fave extensions don't work anymore.
I just wish that the "update firefox" screen would show if (and which) extensions are compatible with the version they want you to upgrade to.
My unsuccessful dist-upgrade notwithstanding, the progress from Debian Etch, through Lenny and now Squeeze is huge - and really nice to see.
Even though Ubuntu is slagged for taking credit for the work that Debian is doing, as much as Ubuntu stands on the shoulders of Debian, in turn Debian stands on the shoulders of all the upstream projects that make up a usable Linux distribution.
None of the full-OS projects/distributions, be they based on Linux or a BSD, can stand alone - it's a huge, interconnected world of projects.
Even though Ubuntu is slagged for taking credit for the work that Debian is doing, as much as Ubuntu stands on the shoulders of Debian, in turn Debian stands on the shoulders of all the upstream projects that make up a usable Linux distribution.
None of the full-OS projects/distributions, be they based on Linux or a BSD, can stand alone - it's a huge, interconnected world of projects.
Quoting:With #3 above, I think it's pretty obvious that Ubuntu's QA is horrible if people have to do this. Imagine if, when you bought a car, you didn't drive said car until later after "all the recalls and bugs were fixed". Would it be practical to drive that car? In the same vein, it is not practical to 'drive' the latest Ubuntu release? To me this is unacceptable. If I had bought a car like this, I'd return it and use something else. The fact that a majority of users out there _DON'T_ use something else means they are either ultimately stupid or a glutton for punishment.
Wait, what?!? That's a terrible analogy! If we must use car analogies, it's more like this: "imagine someone gave you an awesome car for free, with leather seats and Bluetooth, that worked well, even better than the USD15000 car you could go buy from the dealer, but the engine shut off any time you tried to drive on street names that began with the letter 'z'".
I can't believe so many people here are looking the gift horse in the mouth.
Gift horse or Trojan Horse?
Seriously, there are a ton of distros out there that just plain do a better job than Ubuntu, including ones aimed at newbies. Ubuntu really and truly is no big deal and it's high time the community make people aware that it is simply one choice in many.
Seriously, there are a ton of distros out there that just plain do a better job than Ubuntu, including ones aimed at newbies. Ubuntu really and truly is no big deal and it's high time the community make people aware that it is simply one choice in many.
I still recommend Ubuntu to newbies, for one simple reason: googleability. If they have a problem they can very easily search for it in google and usually get back plenty of results to help them out. It makes it easier for new users to help themselves.
Quoting:I still recommend Ubuntu to newbies, for one simple reason: googleability.Fedora is good for that, too.
True, but Fedora is a bit bleeding edge for my taste.
> True, but Fedora is a bit bleeding edge for my taste.
Personally, I consider that somewhat of an understatement. :)
Though pretty much anything which applies to Ubuntu should also work with Mint, so you could always recommend it.
Personally, I consider that somewhat of an understatement. :)
Though pretty much anything which applies to Ubuntu should also work with Mint, so you could always recommend it.
I strongly recommend Mint over Ubuntu as well. It's just better. And yes, almost all of the Ubuntu fixes also apply to Mint.
Quoting:> True, but Fedora is a bit bleeding edge for my taste.And then there's Slackware{,64}-current.
Personally, I consider that somewhat of an understatement. :)
> And then there's Slackware{,64}-current.
Well, yeah. But it works. :)
Well, yeah. But it works. :)
Quoting:Well, yeah. But it works. :)Usually, but there is no promise of having heavily tested updates. Right now, the shipping 32-bit huge kernels are having trouble loading the sunrpc, auth_rpcgss, and nfsd kernel modules, complaining about "invalid module format". Patrick V. is working on a fix before the next -current update.
Fedora ... bleeding edge ... I need to check for major hemorrhaging I guess. Of all the things I have called fedora, bleeding edge has never been one of them.
Thanks Caitlyn, I'm all happy now!
About time based / ready based releases: It's not all that important.
Debian release management is pretty bad when it comes to communication: http://lxer.com/module/newswire/view/116126/#debian which leads to part of it not being ready on time. It _ought_ to release whenever everything is ready or not at all, but for i18n that surely wasn't the case at Valentine's day '09, mainly because of communication / planning problems. Apart from the fact it was postponed for months. Not fixing bugs is not acceptable, but I'm not sure postponing software for months on the other hand is.
OpenBSD has the same release scheme as Ubuntu, and still manages to bring upgrades which don't break stuff as far as I understood. And if it does break stuff, it's at least well documented how to proceed fixing it.
No matter if you run Ubuntu, Debian, OpenBSD, or buy an iPad/PS3, 'waiting a while before buying it' is always a good thing. For Debian because a few days after release whole translation is done, for Ubuntu for the mentioned reasons, for OpenBSD because there will be more discussion about upgrading issues on the web, for iTab/PS3 because the first models overheated, and newer models didn't have this issue, and when it came to the PS3 were even cheaper. Just to show it's not a 'software-only' problem.
Even after products are tested (where I worked, we simulated 7 years of use within a few weeks, and a 'small' test market followed), unforeseen problems might still occur after market introduction, and these should be fixed for the following releases.
About time based / ready based releases: It's not all that important.
Debian release management is pretty bad when it comes to communication: http://lxer.com/module/newswire/view/116126/#debian which leads to part of it not being ready on time. It _ought_ to release whenever everything is ready or not at all, but for i18n that surely wasn't the case at Valentine's day '09, mainly because of communication / planning problems. Apart from the fact it was postponed for months. Not fixing bugs is not acceptable, but I'm not sure postponing software for months on the other hand is.
OpenBSD has the same release scheme as Ubuntu, and still manages to bring upgrades which don't break stuff as far as I understood. And if it does break stuff, it's at least well documented how to proceed fixing it.
No matter if you run Ubuntu, Debian, OpenBSD, or buy an iPad/PS3, 'waiting a while before buying it' is always a good thing. For Debian because a few days after release whole translation is done, for Ubuntu for the mentioned reasons, for OpenBSD because there will be more discussion about upgrading issues on the web, for iTab/PS3 because the first models overheated, and newer models didn't have this issue, and when it came to the PS3 were even cheaper. Just to show it's not a 'software-only' problem.
Even after products are tested (where I worked, we simulated 7 years of use within a few weeks, and a 'small' test market followed), unforeseen problems might still occur after market introduction, and these should be fixed for the following releases.
AZ: You gotta be kidding. Fedora has much more bleeding edge than Gentoo! Newer kernels, it ships with Xorg 1.8 which even isn't in portage yet, it had Nouveau as standard driver before I could even find documentation about how to do it on Gentoo, you get the idea.
I strongly suggest you make the necessary checks as mentioned.
I strongly suggest you make the necessary checks as mentioned.
> Usually, but there is no promise of having heavily tested updates.
True. Running current is like running rawhide or cooker. You're going to get bitten eventually. Unless there's a good reason, I don't bother. Heck, I'm still on 12.2, for various reasons (one being I expect to have a newer machine to try 13.0 out on shortly).
True. Running current is like running rawhide or cooker. You're going to get bitten eventually. Unless there's a good reason, I don't bother. Heck, I'm still on 12.2, for various reasons (one being I expect to have a newer machine to try 13.0 out on shortly).
Quoting:Wait, what?!? That's a terrible analogy! If we must use car analogies, it's more like this: "imagine someone gave you an awesome car for free, with leather seats and Bluetooth, that worked well, even better than the USD15000 car you could go buy from the dealer, but the engine shut off any time you tried to drive on street names that began with the letter 'z'".
You're right, I should have made it clear the 'selection' of a model of car was the idea here not the money.
Post a comment
A free account is needed to post. Write your comment here first: your text is kept in this browser while you sign up or log in, and is waiting for you when you come back.
Your message:Markdown: **bold**, *italic*, [text](https://link), - lists, > quotes, $x^2$ for math. Formatting help
Sign up and post (it is free) or Log in and post
If everything on the install image - say in the default desktop - meets this standard, than that is very powerful indeed.
I wonder how Debian deals with the whole of the distribution - the 20,000+ packages that aren't part of the default installation.
However much we might be uncomfortable with a Stable release only doing security patches and not adding features or even fixing bugs is the very real issue of regressions, which we've all experienced.
So there's a good reason for the whole concept of a stable release.
What I wonder is the relative importance of this method for desktop as opposed to server.
I don't know what sysadmins are doing these days, but way back if they wanted to run Apache, they compiled it themselves with exactly the version and features they wanted. Maybe today there is greater use of precompiled packages even on servers. It sure makes installation and patching easier.
Not that desktops aren't as mission-critical, and especially in enterprise/business uses there are applications and tasks that must keep working and are coded in such a way that updated packages elsewhere in the system can make them perform poorly or not at all.
I'm having just such a problem with Firefox 3.6.x - I have a Web-based app that works with 3.5.x but not 3.6.x. And since the developers really only care about IE, there's probably not going to be a fix. So when I was running FreeBSD 7.3-release last week, having a choice of 3.6.x and 3.5.x (and even 3.0.x) in the package repository was both helpful and a way around problems with regressions in newer versions of the package.