System category

Update-manager just shows a list. I have no way of knowing which repositories are behind which updates.

What is to prevent a third-party repository from pushing updates for *any* package on my system?

Example: Say I add the repo for Banshee media player. I expect (and trust) it will only push updates that involve Banshee, but what is to stop them from pushing a non-banshee related update? Maybe it's a bit paranoid of me, but I'd like to know which updates are coming from which repo. Then, at least if I see a third-party repo pushing an update for something like sudo, then I'll know something is fishy - but right now I have no way of knowing where the updates are coming from, and that bothers me.

Also, the update-mananger, could possibly group the update list by repository.

It's missing a tool or tool feature could display with some precision what an how far is taking ram memory space (physical ram, and swap partition), just like what gmemusage does from MaXX Desktop.

http://img370.imageshack.us/img370/1926/maxxdesktop01gq8.png

Ok, this is an idea for a middleware layer that would act as a "least common denominator" for a GUI that could be Pythonbased, webbbased or whatever.

We need to create a common framework that we use to set up new services, configure services, users, scheduling and you-name-it.

Ok, my example uses Java-syntax for clarity sake but I suggest it to be modeled en UML2 (Use Netbeans 6.1 for example) and then exported to Python (Won't be too hard as it is all XML). Why Python? Many Ubuntu-tools are already build on Python and some services (such as Samba) has Python hooks ready.

Imagine that you could add a new domain in bind using the following syntax:

Bind bind = new Bind();
Zone abertura = new Zone("abertura");
Record www = new Record(CNAME,"www","dns1.openforce.se")
abertura.addRecord(www);
bind.addZone(abertura);
bind.restart();

..the framework adds the needed changes to /etc/named/named.conf.local, creates the files in /var/cache/bind/, restarts the service etc..

On top of our layer we add authenticated XML-RPC - that would make it more generic and we could build GUIs using JavaScript, Python, Java, Perl, Mono, you-name-it.

We could build these wrappers around everything we want to manage: Samba, Cron, MySQL, Postgresql, Postfix, Dovecot.

We use XSL-stylesheets in order to generate the configfiles.


[....]

Kubuntu, a second class distro, no more! (See http://brainstorm.ubuntu.com/idea/478/ and https://bugs.launchpad.net/bugs/150333 for more information).

I propose to create a website to donate money for one more paid developer for Kubuntu (because Canonical has 2 paid GNOME developers, but only has 1 paid KDE developer):
1.) It stipulates the amount of money required to pay a developer for a year.
2.) Anyone who wants to donate, donate what he considers. But only will be forced to donate if it reaches the amount indicated in item number one (the the amount of money achieved will be showing daily on the website).
3.) Perform this idea once a year.

My idea is similar to this and it worked: http://www.pledgebank.com/nouveaudriver

As in linux many application can reside in memory (such as amarok, pidgin, screenlets and ...) it would be very handy to be able to switch among them as we switch between currently active applications.

The routine check should "simply" continue at these position where it has been skipped by user last time.

e.g. store collected data to a status log file so a modified e2fschk can use the data for continuing his work.

Whenever we start the Synaptic package manager, we see many programs available for installation.

However, the Kubuntu (KDE) ones are available in Ubuntu (Gnome).

SPM should be smart enough to remove the KDE repositories from Gnome Ubuntu list, and vice-versa.

Update: Add the option to filter out KDE apps.

I want Access Control List support.

Screenhots of nice GUIs:

http://img521.imageshack.us/my.php?image=nautiluszugriffslistemx8.png

http://www.alobbs.com/albums/albun26/ZFS_acl_dialog1.jpg

May replace the permission tab in nautilus with a nice acl gui.

Most of people know the 'sync' issue is because performance.

The people used to not knowing 'sync' command (specially newbies) used to loose important files regularly. It's a fact, sadly.

Would be interesting some system tool, like a Gnome applet or something like, showing us how far clicking on the 'gksudo sync' custom Gnome Panel launcher is needed.

People tried to install or update to Interpid Ibex may felt surprised suddenly how painful is running Gnome on a machine with 512mb or ram or less. (i'm now using LXDE by default)

Since no much significant features were implemented, it seems to be still dirty code on Gnome should be cleaned to optimize it for computers not so plenty of gigabytes of RAM.

Interesting to see how we can save more than 100mb of ram using LXDE instead of Gnome, even running all Gnome applications, like nautilus, gnome-terminal, gedit, etc.

The linux security model currently fails with interpreters (programs that read scripts such as Perl, bash, python) because it is still possible to run such applications even if they do not have +x (executable) permissions. You simply run them as:
perl program.pl instead of ./program.pl

Therefore, any distro that gives any access to such programs to users, cant prevent users running their own applications in their home directory. Because the user can simply use an interpreted application. Also, its a security risk, because it also means servers could potentially execute such files even if the directory has no +x support. Some might argue that we can protect against that by doing -R, but you honestly cant prevent users reading their own files (it would be silly)

I propose that we add an option to the individual interpreters config files that when enabled, will not allow files which don't have executable access to run.

Otherwise -x offers little security at all. If we are going to use permissions, they should work properly, especially since we include access to these powerful compile-on-the-fly interpreters by default!

in a tree becouse we'll know what to update:

for example:

>compiz
---compiz-core
---compiz-plugin
---compiz-gnome
>evolution
---evolution-data-center
---evolution-webcal
---evolution-plugin
>firefox
---firefox-3.0
---firefox-3.0-gnome-support

etc, update manager should be more intelligent in general!

Im guessing that since Ibex is in beta the background wont be changed, and from what i've seen so far the response has been nothing but negative towards it.

There should be a way for the community to force last minute changes in case something like this were to happen again, where a easily fixable decision went horribly wrong and received tons of bad press....

eg:

http://www.linuxloop.com/news/2008/10/01/sometimes-things-just-dont-work-annouc ing-ubuntus-new-theme-pile-of-crap/

http://opensource.idg.se/2.1014/1.184032/communityn-ubuntus-nya-bakgrundsbild-s uger

http://www.nabble.com/Intrepid-Ibex-Theme-td18931241.html

Im using the artwork as an example, but this should apply to any easily fixable problem, another example could be choice of a certain package over another that would cause an outrage in the community.

I propose that we 'ionice' (adjust the disk I/O priority) for Update Manager, Synaptic, File Roller, apt-get, trackerd and file copy/move operations.

If you run Update Manager, you might have noticed that while it is installing software on your system, doing stuff such as Firefox may be slow on the computer.

Or when you copy a large file, then Firefox may be slower at loading/writing data.

We should lower disk I/O priority for background processes and non-interactive stuff so that they don't slow down interactive applications used by the user.

This would make the system feel more responsive to the user.

Mouse pointer acceleration in current X.org has some issues, that I guess many people do not notice as much as others, so it has not been discussed before (not that I can find anyway). The reason for this is I guess simply that you can get used to the way X.org handles the mouse pointer pretty well. That in itself shouldn't mean the behaviour could not be improved.

Please take a quick look at this X.org documentation that describes the newly developed predictable pointer acceleration:
http://www.x.org/wiki/Development/Documentation/PointerAcceleration

Generally the idea is to improve the feel, accuracy and responsiveness of the mouse pointer.

This I think should be a prime feature for one of the next (K/X)ubuntu releases. So the idea is not so much about the *if* of predictable pointer acceleration, which will be part of X.org in the future, but about the *when* and *how*.

At least one other brainstorm entry (Idea #6567) discusses part of what the predictable pointer acceleration addresses, so there seems to be some demand.

The new configuration options could be implemented directly within system-settings (and similar) to have direct access to and information about the improved feature via the GUI.
Additionally a good default profile could be worked out (if what X.org comes with isn't optimal).
The new behaviour should follow the ideas of the original implementation in the regard that it gives a better feel and accuracy of the mouse pointer and does not add any additional delay or feels unnatural.

without mounting them everytime!

Please add tools for checking and repairing file systems like chkdsk /f in windows.

Mono 2.0 is out, but Intrepid is about to be released with 1.9.1. There are many new and long awaited features on 2.0, including support for almost everything from .NET 2.0 and LINQ support. It would be good if we could base the new version on this release.

The trash bin currently is like a normal folder but with a different icon and the deleted files go to it when deleted them.
But the trash bin should be an isolated zone, any files in that zone must be labeled as dead file, it shouldn't work and can't be effective in anyway in or out the trash; when we press on a file the trash bin should give us the only properties for the file not make it play or work.

at the moment when you double click a file it wil use what ever default application is set to open it


you can right click the file and there will be other programs listed


but you cant do a right click - set default application load - to set what ever default app you want to load the files with



major bummer here


if you vote this down would you be so kind as to explain why