In-Depth Roadmap Analysis For Ubuntu Hardy Heron 8.04

Update: Read the new version of this article revisiting all the ideas and current progress!

The Hardy Heron Roadmap has over 130 new ideas that have been proposed thus far. I’ve examined each one of these ideas in detail, threw out the ones that weren’t interesting to me, wrote an explanation for each, and sorted the list into three categories:

  1. User Experience,
  2. Networking and Security,
  3. Support

Of course, the juiciest items have already made headlines, such as the new theme that is being planned… “I for one, welcome our new hopefully non-brown themed overlords!”

What about the proposed changes that don’t deserve their own headline? Either you have to muck through that entire list, or you can simply read about it here. You can get more information about specific items by clicking on the name of the proposal, which will take you to a page that has a little bit more information about it. Without more from me, here we go.

User Experience

Ubuntu Theme for 8.04
We’re getting a new theme in the next release? Ubuntu has used close to the same theme since Warty Warthog – about 3 years, The REALLY interesting bit about this is how far this reaches… proposed changes include changes to the installer, bootup screen, wallpaper, application splash screens like Gimp and OO.org, Compiz effects, cursors, and even skinning WINE! This is going to be quite a bit of work, but in the end, well worth it.

Third Party Apt
Adding a 3rd party apt repository is more difficult than needed. Editing sources.list, or adding it to the software repository via GUI is a pain. This proposal focuses on creating a standardized file format (let’s call it .install file) which would allow a user to double click on the .install file, then apt would automatically install the program, add the 3rd party repository to the sources.list and automatically manage updates, etc. Very slick idea.

X.Org 7.3
This is something that SHOULD have been included in Gutsy Gibbon, but wasn’t quite ready… ah well, I guess we need to wait 6 months for this. The biggest feature of 7.3 are: Bullet Proof X – The driving force for this is to never boot up into text-mode if something is screwed up with the xorg.conf file. Very nifty. The direction this is headed is to make xorg.conf obsolete, and eventually run without it, which would be fantastic for new users.

Automatix-Ubuntu Team Collaboration
This is certainly a little bit controversial, since Automatix was raked over the coals for breaking upgrades to Feisty Fawn, and now they want to collaborate? Automatix makes stupid easy to install things that are in legal grey areas, such as codecs for playing commercial DVDs, or other proprietary software like Skype, Google Earth, etc. This idea would make it so that Canonical/Ubuntu does not have legal repercussions for including proprietary codecs or other software that isn’t redistributable.

Dual/Multi Monitor Gnome Configuration
Along with X.org 7.3, multiple monitor configuration within a GUI is going to be pretty awesome.

SlickBoot
This proposed change will give the distribution an improved user experience when booting up and shutting down. If you’ve ever seen a Mac boot up, you know exactly what this is trying to emulate. Transitions from the three graphical modes (text, SVGA, and high-resolution) are not really bad at the moment, but if Ubuntu can emulate how a Mac boots up, that would be pretty damn terrific.

Install on an existing filesystem without overwriting /home
This is by far one of the most annoying things that I have to deal with on a regular basis. Personally, I reinstall quite often to test out new distributions. Since getting an external 320GB drive it hasn’t been quite as painful, but for most people, they might need to backup the /home directory to avoid data loss when given the chance to upgrade every 6 months.

Improve Handling of Full Disks
Have you ever run out of hard drive space in Linux? Let me tell you, it is NOT pretty! In some cases, you can end up booting into text mode, with a read-only root partition. It is a royal pain in the ass, to say the least. This proposal is in the early stages, but suggests adding a notification that the disk is almost full, prompting you to run a cleanup wizard, and other ideas that make the problem less likely to happen in the first place, as well as (hopefully) making it easier to recover from if it does happen.

Integrate Prefetch into Bootup
Google’s 2007 Summer of Code is adding hard disk prefetching and optimization resulting in faster bootup times for Hardy Heron.

Easy File Sharing
Make sharing files between Ubuntu machines on a LAN/WLAN easier. Duh!

Single Click Install
Enable easier installation of software from the internet.

Add Remove Software Improvements
This proposes an update to the add/remove software program, which changes it into a sort of “online storefront” where the initial screen shows screenshots, top-rated or brand new applications, etc. If this is approved and pulled off properly, could be a HUGE improvement.

Simple Samba Integration
The Shared Folders utility is getting some upgrades. It will prompt the user to install Samba, no more adding users from the command line, and a simple checkbox will share all home directories on the computer.

Redesign About Ubuntu
“People are used to “About Name of Program” showing a window that gives the software version details, and copyright info. […] Ubuntu should be just as polished.” uname –r always worked just fine for me, but I can see why it’s important.

Networking / Security

Modular /etc/network/interfaces
“Split out the configuration from /etc/network/interfaces into one file per (logical) interface.” This is not a great idea, it’s a pain in the ass already to edit this and maintain correct syntax, why make it so that there are multiple files?

Dial Up Support
This will make setting up and managing Dialup and ASDL out-of-the-box, using Network Manager, very easy.

Improve support for encrypted file systems
Starting with Gutsy Gibbon, you can now install Ubuntu to an encrypted disk. The problem this addresses is adding support to install to a disk which would be auto-partitioned as an encrypted file system from the GUI, instead of using the text-mode alternate installer. Other ideas are to look for key files on USB sticks and other media, instead of just using a password.

Live CD- share this
Direct from the wiki: “Netboot server for easy setup of thin clients and machines which don’t have a CD drive.” Nice idea!

killall gksudo: Stop running GTK as root!
gksudo runs hundreds of thousands of lines of code just to show that that little box that asks for a password when trying to change administrative settings… and it is a little confusing to ask for the password before any changes are committed – why not stop asking for the password until the configuration tool tries to write to a file, then do it in a manner that doesn’t need to utilize as much code to speed things up a little bit.

Support

Measuring Download/Installation Success
In the next release, there may be some big-brother-ish ideas being included. It’s more accurate reporting that can give developers solid numbers for things like how many people download, install, and are able to run Ubuntu. The reason is to find faults in the existing procedures for acquiring Ubuntu, and address where problems arise. Hopefully these numbers will be shared with the community. One area I can already say is a significant problem is the lack of an integrated CD burner for Windows. I’ve heard many users say they’ve burned a copy of a distribution and then it didn’t boot. The primary reason for this is that they simply burned the ISO file itself to the CD, instead of burning the image properly. Now that Ubuntu is targeting a larger audience, it would be neat to have a Windows user simply download an EXE file, open it up, and it prompts to insert a new CD and then provides some additional instructions for getting it installed and setup, perhaps even give an option to print out a PDF of those instructions? The executable would also provide additional information that this proposal is suggesting be recorded.

Identifying Local Users and Groups
This would be really slick. After you put in your location( integrated into the Time Zone selection), during the installation (or post-install) you would be presented with information regarding your local LoCo team and other local resources, such as a Linux User Group in your area. I’ve never been to a LUG or LoCo meeting, perhaps because I needed to seek them out… getting a prompt every reinstall might be annoying, but it would be a good reminder that they exist.

Packaging Section On The Forum
There’s a hidden packaging section (http://ubuntuforums.org/forumdisplay.php?f=44) on the Ubuntu forums that needs to be better utilized. It is not a very active section in the forum, it either needs to be added to the main list of forums and promoted, or simply deleted.

Make screencasts available on the desktop
A Help->Tutorial Videos browser menu? This is a fantastic idea!

Forum Content Certification
Sometimes instructions provided in the forums just don’t work. Often, I’ll find a howto on installing some software for an old version of Ubuntu, and this would create a new forum team that would check the validity of instructions, approve the post, promote the information in the forums, wiki, and elsewhere, and finally involve re-checking the posts after new releases.

Getting teacher input to shape our education offering
This one implies that Edubuntu does not specifically solicit input from teachers, nor do users have a mailing list, forum, etc to voice their complaints. If this is the case, definitely a good idea to reach out and get feedback and suggestions from the people who actually use the OS, right?

Automatic bug reporting
This is certainly an interesting idea, but due to privacy concerns I don’t really want to see this make it into the release. Why emulate Windows here? It can be REALLY annoying to get MORE error popups when things crash.

Ubuntu Mobile Browser
The Ubuntu Mobile and Embedded Project proposes to add a Firefox-based browser which would be completely reworked from the ground up. I’m personally very interested in this, the following quote gave me chills: “Ideally we would get a solution that would be embraced by the Mozilla community and eventually adopted as the “Firefox Mobile” solution.”

Of course, I saved the best for last….

Installing Ubuntu from within Windows
ubuntu-install.exe… nuff said… might not make it though… it is listed as ‘dangling’ which means it can’t be scheduled or has circular dependancies… no idea why it can’t be made to work.

Which ideas do you want to see make it into Hardy Heron? Do you disagree with any of these I’ve listed? Got more to add?

  • http://happylinuxthoughts.blogspot.com happy linux guy

    Just a quick note.
    I upgraded from feisty to gutsy. For some reason my video driver crapped out and wouldn’t work, but it never died and went to the command line. (which actually made it more difficult for me) It defaulted back to a 640×480 mode and went on with business. After trying to reinstall nvidia drivers and messing with it for quite some time, I deleted the xorg.conf file completely and restarted the xorg server. Only then did it use vesa with 1280×1024 resolution. It didn’t rewrite the xorg.conf though. It wasn’t there at all, yet it went into a very useable graphic mode. I’d say it’s implemented quite well, except for the horrendous 640×480 mode. To actually get the nvidia driver working, I reinstalled fresh, and it worked perfect.

  • http://www.livinginx.com LivingInX

    Nice post as usual. Lately, it seems that I have been getting a lot of my Ubunutu news from you. Thanks Wayne.

    @Happy:
    I tried an upgrade at first too. It pretty much borked everything. Reinstall made it nice and smooth.

  • http://www.fsckin.com/ Wayne

    Nice tip, if I ever have that kind of problem again I know what to do.

  • http://technical-itch.co.uk/ Dean @ Technical Itch

    This looks like a great feature set.

    3 features catch my eye:
    Third party apt
    Automatix collaboration
    Install on existing system without overwriting /home

    I tend to mess around installing new distros and 3rd party apps so this looks promising.

  • http://weblogs.mozillazine.org/mitchell/archives/2007/10/beyond_sustainability.html Udo

    Hmm,

    i think Ubuntu should try to come up with an easy backup solution like time machine. Something like that is such an important feature.
    Everything else listed here are just minor little tweaks compared to an easy backup solution that “everybody” can/would use.

  • http://weblogs.mozillazine.org/mitchell/archives/2007/10/beyond_sustainability.html Udo

    2 x Hmm,

    i think Ubuntu should try to come up with an easy backup solution like time machine. Something like that is such an important feature.
    Everything else listed here are just minor little tweaks compared to an easy backup solution that “everybody” can/would use.

  • http://www.SNCSafeNet.com jimbo78255

    While not exactly an UBUNTU issue, it is part of the UBUNTU community … I’d like an UBUNTU colinux product similar to topologilinux. one that is ISO-downloadable, reliable, and works.

    Integrated hamachi at the service level would be a real time saver, also.

    Cheers to all from here in San Antonio!

  • MKx

    RE: Install on existing system without overwriting /home

    I’m pretty novice in Linux, but is it not possible for the installer to remove all directories in root except home and install Ubuntu – maybe with fsck run?

  • http://www.worlds-smartest-man.com Smartest Man in the world

    Cheers from Colorado. Great post!

  • Jimbo

    I think that getting network authentication into the feature list is a MUST. If Ubuntu is going to make any inroads in business it needs to be able to authenticate to a network server.

    This can be done currently but it is no easy installation/configuration

  • Paul

    I don’t understand what keeps people from creating a separate /home partition, or why the installer defaults to putting everything in 1 partition. It allocates a swap partition, whats the big deal in creating /home separately. Then, upon reinstall, don’t format it and mount it as /home.

    The average user doesn’t reinstall their machine constantly, if ever. The user who does reinstall a lot and try different distros should learn how to use the partitioning tool.

  • qbix

    I hope they implement proper fakeraid support. I want to get Ubuntu on my machine, but my Linux experience is not vast. Definitely not enough to feel comfortable attempting the fakeraid guides out there.

  • Ally

    ‘killall gksudo: Stop running GTK as root!’

    I can see your point, but I disagree. I think that rather than completing, say, a big, dirty form/process the user will then be confronted/rejected by it.

    I understand that there’s a lot of code involved just to bring up the initial prompt, however I think this is an issue that requires slimming down gksu, rather than change its method. Just my 2 pence…

  • http://triplefault.org josh

    Improved bluetooth support!

    Getting devices such as bluetooth mice and keyboards setup is quite a pain and involved several command line interactions and all this is often lost over boots or suspend/hibernates and resumes.

    What would be great is an interface to detect and manage bluetooth devices and help to automate setting these up and detecting when they come and go.

  • mikshir

    Good idea about the network install and the install from windows options. However, if possible, I think it would make a ton of people happy if 2 additional install options were directly and fully supported:

    1) Install to USB hard drive (that at least guides the process)
    2) LiveCD on USB flash images (with setting savings feature)

    There are of course how-to’s available but they seem to change just a bit with each release and require some experimentation… plus they can be complicated, especially if things go wrong.

    Boot from USB drive is becoming common and a lot of folks are hesitant to muck about with their primary OS drive.

  • matthews

    Do keep in mind the main focus of LTS relases are to fix bugs not add features so small tweaks and lots of bug fixes are what is planned.

  • http://osversion.com Dhonn

    Solution to home problem.

    /dev/sda1 as / (sane default)
    /dev/sda2 as swap (sane default)
    /dev/sda3 as /home (rest of disk)

    Do not format /home

    If you choose a username that already exists in the home folder it wont delete it. Settings files and all are intact!

  • hp

    Here two more suggests:

    1. NX/FreeNX: public-key auth – wizard
    2. Kerberos: system wide logins for users – wizard

    Btw: booting and decryption at startup via USB is a great idea.

  • Soren Hansen

    I happen to be the author of this:
    “Split out the configuration from /etc/network/interfaces into one file per (logical) interface.”

    “This is not a great idea, it’s a pain in the ass already to edit this and maintain correct syntax, why make it so that there are multiple files?”

    Did it perhaps occur to you that I proposed this change so that it would be easier to maintain? If your in-depth analysis says this is not the case, could you perhaps be a bit more specific and help make the suggestion better?

  • http://gphoneholic.com Alex

    On installing Ubuntu from inside Windows:
    http://wubi-installer.org/ is pretty nice. Not official, but still easy.

  • Sheldonl

    From an enterprise desktop users point of view there should be an easy way from the GUI to join a windows AD domain and share folders out to other users on the same domain.

  • http://www.realistanew.com Travis Watkins

    The ‘killall gksudo’ specification is not complaining about the gksudo application itself, the complaint is that (for example) synaptic and such are running as root and _that_ is a lot of code running as root that doesn’t need to be. The only parts of synaptic that should use root are changing sources.list, reloading the package list, and installing/removing packages. Having the whole app run as root just makes it that much easier to exploit.

  • http://www.fsckin.com/ Wayne

    @Soren Hansen:

    Thanks for the response, Soren.

    I’ve reordered interfaces on a router running Linux in the past, and with only one file needed to be edited very carefully to get it right, it wasn’t the easiest exercise I’ve ever performed.

    This is what I see proposed – and if I’m wrong, please correct me. Say I’ve got 2 logical interfaces, eth0 and wlan0. The relatively simple /etc/network/interfaces becomes three or four different files?

    Why is this smart? Why is this needed? There’s only one explanation I can see, and that’s for network administrator who has 10 logical and a thousand virtual interfaces on a machine.

    If that single network admin has an easier job at the expense of a thousand different people being confused by the unexpected change to /etc/network/interfaces when their network is down and broken, that’s simply NOT the way to do things as I see it.

    I could imagine making the change for Ubuntu Server, but that is the ONLY place it remotely makes sense.

    The “data center”-centric thinking I see demonstrated by this idea is not conducive to an end user being able to figure out what is what when they see this for the first time.

    -Wayne

  • Anonymous

    .install files would be hidden by default and users will not be able to click on them.

  • Anonymous

    I would like the “Restore items from trash” implemented. It’s just a nice feature to have.

  • travis miller

    I’ve been using linux for 10 years now, and whenever i “upgrade” or change distros i just re-install and have had no problem keeping my /home safe. just put it in its own partition. The problem is that when Ubuntu installs, the default partitioning scheme it picks is a little crazy. Maybe if they fixed that then we wouldn’t need a special “don’t delete /home rule”

  • gman

    Similar to wubi is unetbootin. wubi only creates an image file.
    http://en.wikipedia.org/wiki/UNetbootin

  • HackTheGrind

    I like a lot of those ideas, accept two…

    1. Creating a automatic installer like windows’ “setup.exe” would be a HUGE mistake. It would be a huge security issue, opening the doors for new linux viruses, malware, spyware, etc… Plus, it would ruin the integrity of the repository system. The repository ensures that only legit software gets into your system and that is what should be improved upon…

    2. Creating in executable Ubuntu for Windows is A MEGA FREAKIN’ HUGE ASS BAD IDEA! Windows is full of bugs, it is a resource hog, and it is infinitely insecure! So why run Ubuntu through Windows??? It will only make Linux look bad!

    Open source is going to straighten out companies like Microsoft whether they like it or not. If these huge companies don’t begin to change, the people will continue to write code superior to theirs until finally, there software has been stolen so much that they can’t sustain themselves in the business market any longer and they crumble…

    So the point is, the open source movement should continue to do what it does, and Ubuntu is a huge beacon for the open source movement… It should give NO similarity to ANY Microsoft product, as Microsoft does not deserve such a boast. Instead, code developers should continue to be innovative and leave the Microsofts in the dust…

    blah blah blah

    I digress…

  • Pingback: SearchRoads » nude lara croft In-Depth Roadmap Analysis For Ubuntu Hardy Heron 8.04

  • http://www.ubuntu.com/ Soren Hansen

    > I’ve reordered interfaces on a router running Linux in the past, and with only one file needed to be edited very carefully to get it right, it wasn’t the easiest exercise I’ve ever performed.

    Ubuntu has at least since Dapper made sure that once an interface got its name it would never change again. You can add additional interfaces, remove all of them, shuffle them, and put them back in. It’ll still be named the same. It maintains a mapping between MAC address and the interface name. I don’t know which Linux distro you’ve used, but please realise that network device naming is completely orthogonal to the configuration of said interfaces.

    > This is what I see proposed – and if I’m wrong, please correct me. Say I’ve got 2 logical interfaces, eth0 and wlan0.

    Those are physical interfaces.

    > The relatively simple /etc/network/interfaces becomes three or four different files?

    It’s important to point out that at this point, it’s merely a suggestion. Not only that, it’s also an incomplete one. What is on the wiki (which you appear to have not read, though) is not final, and it’s not even sure it’s going to happen at all.

    Now, to answer your question: It depends on your configuration.

    As the proposal is now, if you had:
    iface eth0 inet dhcp

    That would become a file on its own, called /etc/network/interfaces.d/eth0
    with the following contents:
    family inet
    method dhcp

    That’s it.
    If your ath0 was configured statically:
    iface ath0 inet static
    address 192.168.1.1
    network 192.168.1.0
    broadcast 192.168.1.255
    netmask 255.255.255.0
    gateway 192.168.1.254

    That would become at file on its own, called /etc/network/interfaces.d/ath0 with this contents:
    family inet
    method static
    address 192.168.1.1
    network 192.168.1.0
    broadcast 192.168.1.255
    netmask 255.255.255.0
    gateway 192.168.1.254

    You see? Not really a lot changed.. This is all described on the wiki (linked from the spec page on launchpad)

    > Why is this smart?

    Two reasons:
    a) it’s easier to parse and alter programmatically than the current /etc/network/interfaces.
    b) Messing up the configuration of one interface will not be able to affect the configuration of other interfaces.

    > Why is this needed?

    Several reasons:
    To make editing the network configuration easier for other programs (such as various system configuration frontends and so on) without running much of a risk of messing up the users current config.

    Also, if you should happen to make a mistake in one of the files, it will only break one interface, rather than your entire network configuration. If you’ve ever managed a server with multiple interfaces remotely (say, several thousand miles away), you’ll appreciate this.

    > There’s only one explanation I can see, and that’s for network administrator who has 10 logical and a thousand virtual interfaces on a machine.

    Why would this be any more benificial to him/her than anyone else?

    > If that single network admin has an easier job at the expense of a thousand different people being confused by the unexpected change to /etc/network/interfaces when their network is down and broken, that’s simply NOT the way to do things as I see it.

    I imagine the migration will parse the current network config, write new config using this new syntax, and put a friendly notice in /etc/network/interface telling the user where to look for his network config.

    I don’t believe in the use case, you’re presenting.
    a) Switching to a one-file-per-iface approach will make broken network configuration a lot less common, because the syntax has been slightly simplified, and you can only ruin the config of one iface at a time.
    b) If you’ve managed to break it anyhow, you’ve already located the appropriate file to fix it in (because that’s the one you broke, right?).
    c) If you haven’t the skills to resurrect a broken network configuration, you shouldn’t have had to edit it anyway. There should be configuration frontends for that sort of thing, and this change is proposed with the explicit purpose of making it easier for such frontends to do the right thing without messing up your configuration.

    > I could imagine making the change for Ubuntu Server, but that is the ONLY place it remotely makes sense.

    I disagree. Quite the contrary, actually. From the desktop’s perspective, you ought to be completely agnostic as the technicalities of how your network configuration is stored and effectuated. If this is not the case, it means the tools provided to make this process more user friendly are not good enough. Making this change will make the authors of said tools able to spend more time on getting the fancy things right, and less time worrying about parsing and rewriting your /etc/network/interfaces properly.

    > The “data center”-centric thinking I see demonstrated by this idea is not conducive to an end user being able to figure out what is what when they see this for the first time.

    Again: End users who are not technically inclined enough to understand something as simple as “Your network configuration has moved to /etc/network/interfaces.d” are not likely to have had even the slightest chance of being able to fix /etc/network/interfaces anyway.

  • Rune

    Hey man

    Thx for a very interesting article!!!! Good stuff.

    I think Sorens Hansens idea makes perfect sense, and is a very beautiful structure.

    The discussion is reminding me a about the whole “is-coding-by-convention-a-good-thing?” talk regarding frameworks.

    For me this is a convention i would like the os to make for me!

    But different moves and different grooves :)

    Ciao
    Rune

  • http://ibeentoubuntu.blogspot.com Daeng

    I put in a feature request a few weeks ago to have Hardy use a seperate /home partition set noexec by default. Not only would this make upgrades simpler (part of another proposal I made), but it would increase security by limiting executable viruses and worms to the system area, which is more difficult to get access to, and allow easy reistallation if something in the system were hosed.

    Cheers,

    Daeng Bo
    http://ibeentoubuntu.blogspot.com

  • AC

    8.04 is supposed to be an LTS. So, fix those outstanding bugs and forget adding a bunch of new features that will be broken. –AC

  • http://www.fsckin.com/ Wayne

    Soren:
    >Ubuntu has at least since Dapper made sure that once an interface got its name it would never change again.
    I was not aware of this, but certainly good information. I’m curious here, where is the MAC to interface list kept? What happens with interfaces that currently are not detected with a hard MAC?

    >Those are physical interfaces.
    As soon as I posted my reply I knew I reversed it in my head… and I actually did read the read the proposal in full, including the little bit where it says “Before ranting[...]” and I just did the opposite. Whoops!

    >What is on the wiki (which you appear to have not read, though) is not final, and it’s not even sure it’s going to happen at all.
    I understand that the entire list is a suggestion/idea/proposal and use those terms liberally in the article. That said, they’ve already been through an initial approval process and more likely than not will make it in, as long as they are feasible.

    >You see? Not really a lot changed.
    I see. The way I read it in the wiki last night was a bit confusing. Also I do not beleive that the Migration plan was posted last night. I appreciate you clearing this up.

    >Two reasons[...]
    Both of these make sense.

    >To make editing the network configuration easier for other programs (such as various system configuration frontends and so on) without running much of a risk of messing up the users current config.
    I’ve had an upgrade (not Ubuntu) go ballistic and destroy the interfaces file, adding a few dozen non-printable characters instead of newlines, so again, it makes sense to make it easier for the developers.

    >Why would this be any more benificial to him/her than anyone else?
    I don’t agree with the rationality of adding additional complexity to a file that is usually under 5 lines long for a single user machine. If a fictional administrator has a interfaces file that is 2000 lines long, and often makes mistakes when making changes, I can see the reason.
    Let’s step back to reality for a moment. How many installations have more than 2 interfaces? How difficult is it to parse and alter a file with two interfaces? I don’t think it’s THAT HARD that it requires the change.

    >I imagine the migration will parse the current network config, write new config using this new syntax, and put a friendly notice in /etc/network/interface telling the user where to look for his network config.
    Was this migration section just updated in the last 24 hours with this information? I don’t recall reading it previously.

    >a) Switching to a one-file-per-iface approach will make broken network configuration a lot less common, because the syntax has been slightly simplified, and you can only ruin the config of one iface at a time.
    >b) If you’ve managed to break it anyhow, you’ve already located the appropriate file to fix it in (because that’s the one you broke, right?).
    >c) If you haven’t the skills to resurrect a broken network configuration, you shouldn’t have had to edit it anyway. There should be configuration frontends for that sort of thing, and this change is proposed with the explicit purpose of making it easier for such frontends to do the right thing without messing up your configuration.
    >From the desktop’s perspective, you ought to be completely agnostic as the technicalities of how your network configuration is stored and effectuated. If this is not the case, it means the tools provided to make this process more user friendly are not good enough. Making this change will make the authors of said tools able to spend more time on getting the fancy things right, and less time worrying about parsing and rewriting your /etc/network/interfaces properly.

    All good points. I stand corrected.

  • http://freedomdrive.org stelt

    installing Ubuntu from Windows would be awesome, an important step towards a ‘ubiquitous general free software installer’ as described on http://freedomdrive.org

  • John Bales

    Let’s hope Hardy Heron restores the ability to use USB flash drives. If you upgrade from 7.04 to 7.10, you may find that your USB flash drives cannot be mounted, not even using “mount” at the command line. This problem doesn’t seem to exist for fresh installs, only upgrades. If you’ve stored lots of your important files on USB flash drives and upgrade from 7.04 to 7.10, you’re screwed.

  • http://artie11.com aRTie11

    They need better Theme Support – So many different themes to choose from, so many places to install, they should create a theme browser that goes online and finds the themes and then displays a standardised theme preview, and an easy way to download the theme and use it.

    Think Plus for Windows 98 Easy, the end user is really that dumb, and cool transparent windows and window effects will sell them. Glitter does work.

    Trust me, i’m a website developer, they just eat up flashy shit

  • Pingback: André Gondim » Blog Archive » Ubuntu 8.04 Hardy Heron Coming Soon

  • Pingback: Cthulhu Linux » Blog Archive » A Most Interesting List

  • Hoser

    A ZFS install option would be a great addition… I know its not GPL but a little disclaimer and your set. Then they really could do an OSX timemachine application with ease!!!

  • http://www.limmen.org Ivo Limmen

    The “3rd party APT” and “Single click install” seems a bit overkill; I already can install applications by clicking on a deb file. And installing a deb file does not require a extra APT source.
    I would love the failsafe X, I needed it last month :D.

  • Frank

    Hi Soren,

    we’re running a data center with 40 Debian machines which are running tomcat and apache servers which all listen to their own interface. Our packages install a new interface file when they are installed. We have written our own update-interfaces script which just generates /etc/network/interfaces from the files in interfaces.d.

    The benefit is the same as for all the other *.d directories like /etc/cron.d or /etc/init.d or /etc/default for example. You make it possible for packages to install their own config files and for tools to manage a small piece of information instead of a single big one.

    As Ubuntu is moving towards hiding everything behind a graphical UI anyway its only logical to make the configuration easier for the tool. The normal end user shouldn’t have to mess with this anyway.

    I’m not so sure if you really need a different syntax. One file per interface which are then merged together works pretty well for us.

    Cheers
    Frank

  • MKx

    To follow my earlier post, is there any reason not to be able to leave the /home on the same partition without deleting it when installing?

  • Pingback: Where to now? « Computer Helps and Happenings

  • https://launchpad.net/~crimsun Daniel T Chen

    > 8.04 is supposed to be an LTS. So, fix those
    > outstanding bugs and forget adding a bunch of new
    > features that will be broken. –AC

    Overall, the thrust will be stabilising the components and ensuring LTS-to-LTS upgrade ease. However, there will be significant new versions (e.g., Linux 2.6.24 as a base) of some components – it seems many people are in favour, and backporting features becomes incredibly difficult – and simply bringing in those versions inevitably introduces bugs. We all know that ironing out bugs is non-trivial. If you’re not already involved, feel free to jump in – the water’s fine (start at https://bugs.launchpad.net/+login)!

  • Luigi A Cavallo

    I agree that this article is really interesting. thanks. I’ve wanted many of these answers to what will be available Hardy Heron.

    However I choose to differ (but only in part) with your opinion of Automatic Bug Reporting.

    > Automatic bug reporting
    >
    > This is certainly an interesting idea, but due to
    > privacy concerns I don’t really want to see this make
    > it into the release. Why emulate Windows here? It
    > can be REALLY annoying to get MORE error popups when > things crash.

    Firstly, yes, it is very annoying. So why not do it better than Wind**s? (Imagine that hey? :-))

    For example, similarly to the User Statistics option in the Software Sources configuration utility, Ubuntu’s designers can allow us to elect for automatic bug reports (even on a per app basis, protecting privacy options etc) with a view to borrowing a feature from Wind**s automatic bug reporting that I do like.

    Reporting solutions that may be known to a person that falls victim to a known bug. (Or connects you to an appropriate Community Forum?)

    And while development is underway for this, why not make support for this an open API and promote it to other distros and Open Source projects?

    (Maybe this is already underway, and if so just ignore what I’m saying and forward that info to whomsoever it needs to get to, please. :-))

  • Ray

    One thing missing for ordinary people is Chat client, with video and audio. I recently installed Ubuntu for one of my client and he is computer illiterate, he had no problem using it. But when it came to chatting applications, I had to install XP so he can talk to his family overseas. Now that is a bummer!

  • Ovidiu C.

    I’d like to see an Ubuntu/GoboLinux cross.

    A time machine feature that saves stuff on an external hard drive doesn’t strike me as a very good idea. External drives don’t last very long. Not nearly as long as internal drives. But a easy to use backup application that can save on network drive or (why not) a second internal hard drive would be nice. It doesn’t have to be as flashy as the the OS X Time Machine though. :)

  • Pingback: Prime News Blog » Blog Archive » wonder woman toon porn In-Depth Roadmap Analysis For Ubuntu Hardy Heron 8.04

  • http://edtechdev.blogspot.com/ Doug

    I would suggest

    -an easy backup solution (perhaps improve ‘simple backup’ which doesn’t even work most of the times i’ve tried it)
    -fix nautilus list view columns and drag & drop. Nautilus seems to have regressed from feisty to gutsy.
    -provide better support, right now an irc channel is the top listed support option. An irc channel.
    -better webcam support (3rd party app thing, not ubuntu)
    -how about a shutdown button that doesn’t dance around the top bar from boot to boot
    -fix all the reports gutsy bugs with wireless & suspend support
    -Better support for external displays. Do you know that about the only way to use a projector is to restart gnome or the computer? nvidia-settings crashes the machine.
    -support Wubi. If you really want to increase market share, you need an install that doesn’t require partitioning or even figuring out how to boot to a CD.