Written by Lachu the 10 Sep 09 at 07:55.

Ubuntu(and other distribution) should provide D-BUS based API for installer/uninstallers. You will get extra points if you integrate it with package database ;-) .

It should been integrated with PolicyKit, so administrators can give users access to install many kind of software(example software, which no needs setting suid bit and installing in /opt and don't change any files, can be installed by normal user; software which needs install in /sbin or /bin or /usr/bin, /usr/sbin should be installed only by superuser; or whatever).

This API should gives many functions, like copy executable into system directory(/bin), copy executable into superuser directory(/sbin, /usr/sbin or whatever), install user library(/usr/lib), install system-wide library(/lib), change file(delete existing file or writing own in the same place), change group of file, change owner of file, set uid bit, set guid bit, install nonexecutables(data) to /usr/share/*, installing user software(only have access to /opt), start installation.

All this action will be logged and reported. Additionally user may have simple uninstall software by system tools, because all performed operation are logged.

Installer should first call start installer function, so system show message to user like: This program will install software - continue? If user agree, DBUS returns the path, where the installer should unpack all of the files (in /tmp directory). Of course installer should register software name, etc., but it is long term task. If this is a game, it should install in /opt, so it call DBUS API to copy file from unpack folder into specify /opt folder, like copy_file_to_opt(file_name, opt_suffix).

Ubuntu(and other distribution) should provide D-BUS based API for installer/uninstallers. You will get extra points if you integrate it with package database ;-) . It should been integrated with PolicyKit, so administrators can give users access to install many kind of software(example software, which no needs setting suid bit and installing in /opt and don't change any files, can be installed by normal user; software which needs install in /sbin or /bin or /usr/bin, /usr/sbin should be installed only by superuser; or whatever). This API should gives many functions, like copy executable into system directory(/bin), copy executable into superuser directory(/sbin, /usr/sbin or whatever), install user library(/usr/lib), install system-wide library(/lib), change file(delete existing file or writing own in the same place), change group of file, change owner of file, set uid bit, set guid bit, install nonexecutables(data) to /usr/share/*, installing user software(only have access to /opt), start installation. All this action will be logged and reported. Additionally user may have simple uninstall software by system tools, because all performed operation are logged. Installer should first call start installer function, so system show message to user like: This program will install software - continue? If user agree, DBUS returns the path, where the installer should unpack all of the files (in /tmp directory). Of course installer should register software name, etc., but it is long term task. If this is a game, it should install in /opt, so it call DBUS API to copy file from unpack folder into specify /opt folder, like copy_file_to_opt(file_name, opt_suffix). I know is less secure than installing package, but is more secure than giving installer root privileges.

Written by dandart the 10 Sep 09 at 20:43.

Create a format that all distros can use and is easy to implement. Perhaps integrate into Apt/Rpm or have packages that both can understand somehow. Then we can download "Linux" package without having to worry if we want Apt or Rpm.

Create a format that all distros can use and is easy to implement. Perhaps integrate into Apt/Rpm or have packages that both can understand somehow. Then we can download "Linux" package without having to worry if we want Apt or Rpm.

Written by andruk the 11 Sep 09 at 02:12.

The reason apt needs root permissions is because the applications it installs are installed system-wide and therefore affect other users. If we allowed users to install programs to their home directory, then they could run their installed applications without affecting other users, and therefore wouldn't need to give apt root access.

This would have advantages with home users because my Applications menu wouldn't get cluttered when my daughter installs Barbietown to her home directory.

This would have advantages with sysadmins because they can run a piece of software from their home directory to test it out before installing it system-wide.

The system paths would have to be modified to look for libraries and executables in the home directory, but applications like Xournal already support this. This would also require the Applications menu to look in the user's home directory for the menus as well.

The reason apt needs root permissions is because the applications it installs are installed system-wide and therefore affect other users. If we allowed users to install programs to their home directory, then they could run their installed applications without affecting other users, and therefore wouldn't need to give apt root access. This would have advantages with home users because my Applications menu wouldn't get cluttered when my daughter installs Barbietown to her home directory. This would have advantages with sysadmins because they can run a piece of software from their home directory to test it out before installing it system-wide. The system paths would have to be modified to look for libraries and executables in the home directory, but applications like Xournal already support this. This would also require the Applications menu to look in the user's home directory for the menus as well.

Written by Lachu the 12 Sep 09 at 17:59.

Like in solution #3:
The main problem is many application have hardcoded path. There's no way to detect/guess path of specified file. Of course, we can recompile package, but it lack functionality for users.

The solution is simple. Create script scheme for setting environment, where specify files are placed. It should set PATH, LD_LIBRATY_PATH, XDG_HOME, CONFIG_PATH, etc. Script will also checking dependency and fixing it if not satisfied(also using install into directory technology).

Create also library, which allows program to read these setting/environment. Program should don't read file from /etc/program_config, but from $CONFIG_PATH/program_config.
If some environment isn't set, library should return default value.

User will have many scheme of installing application - install in: user home directory, system-wide(/), opt(/opt), custom paths.

And of course -- allow to set flag SUPPORT_PATH_SELECTING for package.

Like in solution #3: The main problem is many application have hardcoded path. There's no way to detect/guess path of specified file. Of course, we can recompile package, but it lack functionality for users. The solution is simple. Create script scheme for setting environment, where specify files are placed. It should set PATH, LD_LIBRATY_PATH, XDG_HOME, CONFIG_PATH, etc. Script will also checking dependency and fixing it if not satisfied(also using install into directory technology). Create also library, which allows program to read these setting/environment. Program should don't read file from /etc/program_config, but from $CONFIG_PATH/program_config. If some environment isn't set, library should return default value. User will have many scheme of installing application - install in: user home directory, system-wide(/), opt(/opt), custom paths. And of course -- allow to set flag SUPPORT_PATH_SELECTING for package.

Written by o-bin-lad3n the 15 Sep 09 at 20:05.

This is much like #2, but why start from scratch?
We could push for better support and integration of autopackage with apt, rpm, etc.
Also, I think the need for root-permissions itself is not the real problem, it's the need to blindly trust the file you just downloaded. Of course, there is always a way to mess with a system, when the user executes your software, but this would be way better than "sudo ./somefileijustdownloaded". Autopackage can run with user rights and install into your home directory. And you can still integrate gpg-keys with this solution.

I'm not suggesting this as THE universal package format, but as a common solution for third party/proprietary software and commercial games.

This is much like #2, but why start from scratch? We could push for better support and integration of autopackage with apt, rpm, etc. Also, I think the need for root-permissions itself is not the real problem, it's the need to blindly trust the file you just downloaded. Of course, there is always a way to mess with a system, when the user executes your software, but this would be way better than "sudo ./somefileijustdownloaded". Autopackage can run with user rights and install into your home directory. And you can still integrate gpg-keys with this solution. I'm not suggesting this as THE universal package format, but as a common solution for third party/proprietary software and commercial games.

Written by darkjavi the 21 Sep 09 at 16:47.

And what about giving to apt/aptitude/synaptic the ability to control/install any software on earth not only deb packages.
apt could run the instalation from is inside and chroot the destination of the files generated by the installer into his own folders, so we can manage all the computer software from a single interface

And what about giving to apt/aptitude/synaptic the ability to control/install any software on earth not only deb packages. apt could run the instalation from is inside and chroot the destination of the files generated by the installer into his own folders, so we can manage all the computer software from a single interface

Written by Lachu the 23 Sep 09 at 17:29.

Create sandbox for installers. It will check selinux context and when it isn't designed for installers, it will change it and fork with exec on executable file! It can't sets UID/GID bits or change any existing file, but can access to system directories.

Create sandbox for installers. It will check selinux context and when it isn't designed for installers, it will change it and fork with exec on executable file! It can't sets UID/GID bits or change any existing file, but can access to system directories.

Written by Afroman10496 the 27 Sep 09 at 02:31.

You all know how Wine works, right: importing an opensource clone of the Windows API. Now, imagine that for the Red Hat YUM/RPM compatibility. It's already opensourced, so you won't have to actually clone it. Also, you already have most of what you need because it's all Linux :>. Just offer it as a regular package and when users click on the .rpm just offer a dialog asking if they want to install it or Alien, showing all the pros and cons.

You all know how Wine works, right: importing an opensource clone of the Windows API. Now, imagine that for the Red Hat YUM/RPM compatibility. It's already opensourced, so you won't have to actually clone it. Also, you already have most of what you need because it's all Linux :>. Just offer it as a regular package and when users click on the .rpm just offer a dialog asking if they want to install it or Alien, showing all the pros and cons.

Written by ratdude747 the 27 Sep 09 at 22:01.

that would fix the rpm issue. beginners seem to get "scared" of the command line, and a gui saves time. that way, no matter whether they download an rpm or a deb, they are covered.

making it the default app for rpm would make it so a beginner wouldn't extract it by accident with archive manager. maybe you could then auto-load the resulting .deb.

that would fix the rpm issue. beginners seem to get "scared" of the command line, and a gui saves time. that way, no matter whether they download an rpm or a deb, they are covered. making it the default app for rpm would make it so a beginner wouldn't extract it by accident with archive manager. maybe you could then auto-load the resulting .deb.

Written by warlock24 the 7 Oct 09 at 17:38.

Do not multiply another installer formats, apis, ect. they only do unnecesary complicity of Ubuntu. Promote DEB packages to become unofficial standard for every linux. Develop visual tools for easy and fast creation of deb files (something like debGLADE?).

Do not multiply another installer formats, apis, ect. they only do unnecesary complicity of Ubuntu. Promote DEB packages to become unofficial standard for every linux. Develop visual tools for easy and fast creation of deb files (something like debGLADE?).

Written by Lachu the 28 Oct 09 at 07:42.

Not create new package format/packaging system, but allow to pack many packages into single archive. Structure will looks like this:

root
|
+definition&scripts
| |
| +deb
| +rpm
| .....
|
+files(must have structure like tar.gz - if in specify distribution one file must be moved into other directory, it's task for script)
|
+common definitions (name of author, license, features - installation into directory, etc.)


How distribution should handle it? In simple way. It will only unpack it into /tmp and handle as normal package(ex. run installation script for this distribution). It can also generate .deb or .rpm.

It should helps users, who like to download software from network. It can also helps game creators, because CD's can be used as repository of software.

Not create new package format/packaging system, but allow to pack many packages into single archive. Structure will looks like this: root | +definition&scripts | | | +deb | +rpm | ..... | +files(must have structure like tar.gz - if in specify distribution one file must be moved into other directory, it's task for script) | +common definitions (name of author, license, features - installation into directory, etc.) How distribution should handle it? In simple way. It will only unpack it into /tmp and handle as normal package(ex. run installation script for this distribution). It can also generate .deb or .rpm. It should helps users, who like to download software from network. It can also helps game creators, because CD's can be used as repository of software. Maybe there can occurs some problems with binaries, but most data of program is a resources, like icons, graphics, etc. Resources are shared, so there is a reason to create hybrid format.

Written by ivo000 the 1 Mar 11 at 20:11.

Integrate Autopackage package installer in Ubuntu Software Center

Integrate Autopackage package installer in Ubuntu Software Center

Written by SpamBadger the 29 Feb 08 at 00:17.

Ubuntu Brainstorm was updated in January 2009. Since the idea #288 was submitted before this update, its rationale and solution are not separated. Please vote accordingly, and if you have the necessary rights, please separate the rationale from the solution. Thanks!

<i>Ubuntu Brainstorm was updated in January 2009. Since the idea #288 was submitted before this update, its rationale and solution are not separated. Please vote accordingly, and if you have the necessary rights, please separate the rationale from the solution.</i><br /> Thanks!

Written by domsmith the 8 Feb 09 at 13:24.

More advanced controls for setup of most common mice.

The ability to remove mouse acceleration to give a true 1:1 pixel movement within xwindows/gnome.

Ability to set binds for mice with more than 3 buttons via the GUI.

Auto detection of mouse dpi 400/800/1600/2000 etc.

Number values instead of the current slider.

More advanced controls for setup of most common mice. The ability to remove mouse acceleration to give a true 1:1 pixel movement within xwindows/gnome. Ability to set binds for mice with more than 3 buttons via the GUI. Auto detection of mouse dpi 400/800/1600/2000 etc. Number values instead of the current slider.

Written by quenturi the 29 Feb 08 at 02:09.

Ubuntu Brainstorm was updated in January 2009. Since the idea #478 was submitted before this update, its rationale and solution are not separated. Please vote accordingly, and if you have the necessary rights, please separate the rationale from the solution. Thanks!

<i>Ubuntu Brainstorm was updated in January 2009. Since the idea #478 was submitted before this update, its rationale and solution are not separated. Please vote accordingly, and if you have the necessary rights, please separate the rationale from the solution.</i><br /> Thanks!

Written by Phase the 29 Feb 08 at 16:03.

Ubuntu Brainstorm was updated in January 2009. Since the idea #1532 was submitted before this update, its rationale and solution are not separated. Please vote accordingly, and if you have the necessary rights, please separate the rationale from the solution. Thanks!

<i>Ubuntu Brainstorm was updated in January 2009. Since the idea #1532 was submitted before this update, its rationale and solution are not separated. Please vote accordingly, and if you have the necessary rights, please separate the rationale from the solution.</i><br /> Thanks!

Written by Vivien the 28 Feb 08 at 16:35.

Ubuntu Brainstorm was updated in January 2009. Since the idea #72 was submitted before this update, its rationale and solution are not separated. Please vote accordingly, and if you have the necessary rights, please separate the rationale from the solution. Thanks!

<i>Ubuntu Brainstorm was updated in January 2009. Since the idea #72 was submitted before this update, its rationale and solution are not separated. Please vote accordingly, and if you have the necessary rights, please separate the rationale from the solution.</i><br /> Thanks!

Written by Aphoxema the 23 Sep 08 at 15:26.

Since mounting and unmounting a filesystem over and over again is less hazardous than removing an unmounted filesystem, wouldn't it be as simple as having USB devices unmount automatically after so many minutes without use?

Unmounting any USB drives after, say, 5 minutes without use (by default that is, the option to change it should be available somewhere) would make accidental unplugging less of a risk. It would protect the user, it wouldn't make it all that difficult to mount the filesystem again when it was needed, and it's just damn easy to implement.

Since mounting and unmounting a filesystem over and over again is less hazardous than removing an unmounted filesystem, wouldn't it be as simple as having USB devices unmount automatically after so many minutes without use? Unmounting any USB drives after, say, 5 minutes without use (by default that is, the option to change it should be available somewhere) would make accidental unplugging less of a risk. It would protect the user, it wouldn't make it all that difficult to mount the filesystem again when it was needed, and it's just damn easy to implement.

Written by borsook the 25 Dec 08 at 23:47.

If I install package A system will propose installing packages B,C,D that A depends upon. But if I uninstall A B,C,D will be left even if nothing else uses them. Of course this can be checked and done manually, but wouldn't it be better if Synaptic checked if anything else depends upon those packages and ask the user if he/she wants to uinstall them too?

If I install package A system will propose installing packages B,C,D that A depends upon. But if I uninstall A B,C,D will be left even if nothing else uses them. Of course this can be checked and done manually, but wouldn't it be better if Synaptic checked if anything else depends upon those packages and ask the user if he/she wants to uinstall them too?

Written by Kosimo the 10 May 08 at 13:16.

Ubuntu Brainstorm was updated in January 2009. Since the idea #8326 was submitted before this update, its rationale and solution are not separated. Please vote accordingly, and if you have the necessary rights, please separate the rationale from the solution. Thanks!

<i>Ubuntu Brainstorm was updated in January 2009. Since the idea #8326 was submitted before this update, its rationale and solution are not separated. Please vote accordingly, and if you have the necessary rights, please separate the rationale from the solution.</i><br /> Thanks!

Written by nick.colgan the 29 Jan 09 at 03:51.

Add a "scan wireless networks" option to right click context menu of nm-applet.

Add a "scan wireless networks" option to right click context menu of nm-applet.

Written by anabelle the 29 Jan 09 at 15:01.

Since you click to see if you are connected or to connect to a new network it could work similar to OSX.

Once you click it scans for networks and include new results in the lists... no right clicking or anything.

Since you click to see if you are connected or to connect to a new network it could work similar to OSX. Once you click it scans for networks and include new results in the lists... no right clicking or anything.

Written by elbel86 the 1 Feb 09 at 04:19.

wicd is a great alternative to network manager and offers many more options in its gui, including a refresh button. Of course, wicd isn't even in the repos yet, so it could take some time to get it in.

wicd is a great alternative to network manager and offers many more options in its gui, including a refresh button. Of course, wicd isn't even in the repos yet, so it could take some time to get it in.

Written by Endolith the 2 Feb 09 at 07:47.

If there is a problem with Network Manager not showing new networks when they have changed, then fix that problem.

We should not be adding "Refresh" buttons to things that should always be up-to-date. When would you *not* want the list refreshed? Never. If Network Manager were functioning optimally, the button would serve no purpose, so there is no reason to add it.

Let's make things function correctly instead of adding poor workarounds.

If there is a problem with Network Manager not showing new networks when they have changed, then fix that problem. We should not be adding "Refresh" buttons to things that should always be up-to-date. When would you *not* want the list refreshed? Never. If Network Manager were functioning optimally, the button would serve no purpose, so there is no reason to add it. Let's make things function correctly instead of adding poor workarounds.

Written by tchalvakspam the 18 Feb 09 at 20:04.

I'm paraphrasing the solution that one of the developers has implemented (I believe in a version slightly after 0.7) for the purpose of clarifying the brainstorm understanding of what the developers have done to solve the problem.

One of the developers has said that more recent patches to NetworkManager will result in the following behavior:

- When a user interacts with the applet, a rescan will occur immediately if one hasn't within the last 20 seconds. For two minutes thereafter, it will do 20 second interval rescans, then drop to 120 second scans.

- When the applet starts up or the wireless card is re-enabled after being disabled, NetworkManager will do 20 second interval scans for two minutes, and drop back down to 120 second interval scans after that.

So by NetworkManager version 0.7.1 or 0.7.3 or so, any other behavior that doesn't get a rescan as fast as that should be some kind of bug.

I'm paraphrasing the solution that one of the developers has implemented (I believe in a version slightly after 0.7) for the purpose of clarifying the brainstorm understanding of what the developers have done to solve the problem. One of the developers has said that more recent patches to NetworkManager will result in the following behavior: - When a user interacts with the applet, a rescan will occur immediately if one hasn't within the last 20 seconds. For two minutes thereafter, it will do 20 second interval rescans, then drop to 120 second scans. - When the applet starts up or the wireless card is re-enabled after being disabled, NetworkManager will do 20 second interval scans for two minutes, and drop back down to 120 second interval scans after that. So by NetworkManager version 0.7.1 or 0.7.3 or so, any other behavior that doesn't get a rescan as fast as that should be some kind of bug.

Written by hemanth.hm the 8 Mar 09 at 18:27.

WiFi Radar is a Python/PyGTK2 utility for managing WiFi profiles.It enables you to scan for available networks and create profiles for your preferred networks. At boot time, running WiFi Radar will automatically scan for an available preferred network and connect to it. You can drag and drop your preferred networks to arrange the profile priority.

WiFi Radar is a Python/PyGTK2 utility for managing WiFi profiles.It enables you to scan for available networks and create profiles for your preferred networks. At boot time, running WiFi Radar will automatically scan for an available preferred network and connect to it. You can drag and drop your preferred networks to arrange the profile priority.

Written by korin43 the 28 Oct 09 at 19:43.

#6 is nice, but it leaves users feeling like NetworkManager isn't rescanning at all. It would be better if it had some sort of feedback like "Scanning for Networks".

#6 is nice, but it leaves users feeling like NetworkManager isn't rescanning at all. It would be better if it had some sort of feedback like "Scanning for Networks".

Written by tz the 4 Dec 09 at 17:27.

20 seconds is longer than it takes to load most complex web pages, yet that is what everyone is stuck with, but it might actually be 2 minutes.

I would like the option to scan every X seconds when the applet is active, and Y seconds when it is in background. If you constantly change APs, the current settings are too long, but if you are always using the same one they are too short.

Some hardware does bad things (e.g. can't receive or transmit) when you initiate a scan, so you might not want it more frequently, but if I have a good adapter I should be able to unlimit it.

20 seconds is longer than it takes to load most complex web pages, yet that is what everyone is stuck with, but it might actually be 2 minutes. I would like the option to scan every X seconds when the applet is active, and Y seconds when it is in background. If you constantly change APs, the current settings are too long, but if you are always using the same one they are too short. Some hardware does bad things (e.g. can't receive or transmit) when you initiate a scan, so you might not want it more frequently, but if I have a good adapter I should be able to unlimit it.

Written by tz the 4 Dec 09 at 17:48.

Most wireless cards can receive beacons without a scan and most APs ("visible") send them a few times per second. The list returned by the scan includes these - but I think there is a call which returns the result WITHOUT DOING AN ACTUAL SCAN (sending query packets, etc.).

For wireless cards which support this feature, the display list should be updated every second from the list of visible beacons without doing any explicit scan.

Most wireless cards can receive beacons without a scan and most APs ("visible") send them a few times per second. The list returned by the scan includes these - but I think there is a call which returns the result WITHOUT DOING AN ACTUAL SCAN (sending query packets, etc.). For wireless cards which support this feature, the display list should be updated every second from the list of visible beacons without doing any explicit scan.

Written by kuluizemp.openconscience the 1 Dec 10 at 13:05.

we should have a simple and easy way of enable just one wireless network to prevent unwanted mistaken wireless connections and also perhabes a cheak list where you check and uncheck on the wireless connections menu what connections a user want to allow

we should have a simple and easy way of enable just one wireless network to prevent unwanted mistaken wireless connections and also perhabes a cheak list where you check and uncheck on the wireless connections menu what connections a user want to allow

Written by Arioch the 28 Feb 08 at 15:26.

I propose to the development team (both Ubuntu and by extension Gnome) to work on the improvement of boot up times in Ubuntu systems.

I propose to the development team (both Ubuntu and by extension Gnome) to work on the improvement of boot up times in Ubuntu systems.

Written by tchalvakspam the 23 Jan 09 at 21:53.

In addition to speeding up the processes involved, which is key, make it simple for the user to streamline their own bootup. Installing different software can -always- slow down a bootup, no matter how fast it was initially.

Thus, to -keep- boot time short for the many different ways that people use ubuntu, provide the ubuntu user with a simple default tool to remove, delay the start of, or uninstall software that take the most time during startup. The combination of a default boot-time analysis tool and the current session startup manager would provide allow easy boot streamlining, and to a lesser extent so would showing on-screen indicators of what things are being started up over the course of the normal ubuntu boot load.
Allowing user analysis is the key feature missing to make streamlining available to the individual in addition to the developer.

In addition to speeding up the processes involved, which is key, make it simple for the user to streamline their own bootup. Installing different software can -always- slow down a bootup, no matter how fast it was initially. Thus, to -keep- boot time short for the many different ways that people use ubuntu, provide the ubuntu user with a simple default tool to remove, delay the start of, or uninstall software that take the most time during startup. The combination of a default boot-time analysis tool and the current session startup manager would provide allow easy boot streamlining, and to a lesser extent so would showing on-screen indicators of what things are being started up over the course of the normal ubuntu boot load. Allowing user analysis is the key feature missing to make streamlining available to the individual in addition to the developer.

Written by aadityabhatia the 29 Jan 09 at 04:57.

Some programs take too long and are not required immediately upon boot. For example, sendmail takes forever to start and a regular desktop user usually wants to see a login screen asap, so sendmail can wait. Such programs could be identified and their initiation can be delayed while rest of the system continues booting up.

Some programs take too long and are not required immediately upon boot. For example, sendmail takes forever to start and a regular desktop user usually wants to see a login screen asap, so sendmail can wait. Such programs could be identified and their initiation can be delayed while rest of the system continues booting up.

Written by aadityabhatia the 29 Jan 09 at 05:04.

There is a contrast in the comments: some machines take more time before login screen, some do after. In many cases it's the compiz and in some it's dbus or a part of gnome.

The idea is to have a profiler start at the beginning of boot process, on demand, and measure the time taken by each program at startup, writing it to a log file. This will help users figure out what part takes the most time, and will simplify profiling the boot process.

Not sure how much effort is involved, but sounds possible. Any takers?

There is a contrast in the comments: some machines take more time before login screen, some do after. In many cases it's the compiz and in some it's dbus or a part of gnome. The idea is to have a profiler start at the beginning of boot process, on demand, and measure the time taken by each program at startup, writing it to a log file. This will help users figure out what part takes the most time, and will simplify profiling the boot process. Not sure how much effort is involved, but sounds possible. Any takers?

Written by Id2ndR the 15 Feb 09 at 16:10.

The programs that are loaded just after the login are always the same : gnome-session, gnome-daemons etc.

What take time to load the session is to read them from the hard drive. So as we are sure they will be loaded, we can prefetch them.

This can be extended to user's programs just after he give his name to gdm (so while he is entering his password).

The programs that are loaded just after the login are always the same : gnome-session, gnome-daemons etc. What take time to load the session is to read them from the hard drive. So as we are sure they will be loaded, we can prefetch them. This can be extended to user's programs just after he give his name to gdm (so while he is entering his password).

Written by kleytonn the 2 Mar 08 at 12:40.

Turn fast and light the Gnome enviroment with a fast boot. Today, a login in GDM cost 10 or 20 seconds to load a user desktop in one fast computer.
KDE 4.0.1 is new but is visible more light than Gnome.

Turn fast and light the Gnome enviroment with a fast boot. Today, a login in GDM cost 10 or 20 seconds to load a user desktop in one fast computer. KDE 4.0.1 is new but is visible more light than Gnome.

Written by slsolaris the 23 Mar 09 at 17:11.

gnome is too slow, it need a lot of implrovements, like when u see icons overlap any other, gnome panel bar is not real transparent, and i bring this things in mind just because i can not understand how xfce that does have this features is faster and and lighter...

i mean gnome should work in hand with them and give them things that they already do not have and they could give us some very important help to improve things

gnome is too slow, it need a lot of implrovements, like when u see icons overlap any other, gnome panel bar is not real transparent, and i bring this things in mind just because i can not understand how xfce that does have this features is faster and and lighter... i mean gnome should work in hand with them and give them things that they already do not have and they could give us some very important help to improve things

Written by writser the 28 Feb 08 at 16:49.

The following changes could be done

- Merge 'Screensaver', 'Screen Resolution' and 'Screens and Graphics'. -- Already Done.
- Merge 'Network', 'Network Proxy' and 'Network Tools'.
- Merge 'Update Manager, Synaptic Package Manager, Software Sources'.
- Merge 'Encryption and Keyrings', 'Authorizations', 'Keyring Manager'.

The following changes could be done - Merge 'Screensaver', 'Screen Resolution' and 'Screens and Graphics'. -- Already Done. - Merge 'Network', 'Network Proxy' and 'Network Tools'. - Merge 'Update Manager, Synaptic Package Manager, Software Sources'. - Merge 'Encryption and Keyrings', 'Authorizations', 'Keyring Manager'.

Written by yule the 18 Jan 09 at 17:28.

I believe the "Preferences" and "Administration" sub-menus of the "System" menu should be re-structured similarily to "Fedora 10", splitting the items into categories of interest: Look and Feel, Input Devices etc.

Also, but this is probably a generic Gnome defect, items like "Screensaver", "Appearance", "Screen Resolution" should be merged together.

I believe the "Preferences" and "Administration" sub-menus of the "System" menu should be re-structured similarily to "Fedora 10", splitting the items into categories of interest: Look and Feel, Input Devices etc. Also, but this is probably a generic Gnome defect, items like "Screensaver", "Appearance", "Screen Resolution" should be merged together.

Written by jeypeyy the 18 Jan 09 at 21:05.

Put the most important things that a user will need in Preferences and Administration. If the user want to change less important stuff he/she could go to System->'control panel' where the user can change everything that is configurable.

What I mean with important is something like "screen resolution" while "Sun java 6 policy tool" is less important.

Put the most important things that a user will need in Preferences and Administration. If the user want to change less important stuff he/she could go to System->'control panel' where the user can change everything that is configurable. What I mean with important is something like "screen resolution" while "Sun java 6 policy tool" is less important.

Written by c0p3rn1c the 6 Feb 09 at 15:33.

I'm not asking to remove the existing shortcuts in the system menu just to add a shortcut to the gnome-control-center, this makes it much easier for beginners to find what they are looking for. Especially when users are coming from Windows or Mac they are used to a having a control center.

On top of increased similarity to the most popular commercial OS's gnome-controle-center also adds extra functionality with the search box for people who are not used to Ubuntu yet and or are not yet using gnome-do.

Please try the gnome-control-center menu before you vote against it.
It's really a matter of choosing increasing usability over the possible redundancy issue.

I'm not asking to remove the existing shortcuts in the system menu just to add a shortcut to the gnome-control-center, this makes it much easier for beginners to find what they are looking for. Especially when users are coming from Windows or Mac they are used to a having a control center. On top of increased similarity to the most popular commercial OS's gnome-controle-center also adds extra functionality with the search box for people who are not used to Ubuntu yet and or are not yet using gnome-do. Please try the gnome-control-center menu before you vote against it. It's really a matter of choosing increasing usability over the possible redundancy issue.

Written by dael99 the 14 Feb 09 at 00:56.

All the administration task should be done in the Gnome control center, but nowdays it's just a launcher for gnome-conf apps.

The Gnome Control Centre should be a homegenius way of do things, use the same windows for operations and provide a back button, just like a file explorer (more intuitive no?)

Instead of launching a new windows it show be showed (not nested) to the main area of the Centre. This way we could compete with MS, KDE and MAC in terms of a nice control center.


(also we save some space on the gnome pannel)

All the administration task should be done in the Gnome control center, but nowdays it's just a launcher for gnome-conf apps. The Gnome Control Centre should be a homegenius way of do things, use the same windows for operations and provide a back button, just like a file explorer (more intuitive no?) Instead of launching a new windows it show be showed (not nested) to the main area of the Centre. This way we could compete with MS, KDE and MAC in terms of a nice control center. (also we save some space on the gnome pannel)

Written by JimmyVolatile the 10 Feb 09 at 22:48.

Make a nice submenu structure following exactly the category structure of the Gnome Control Center.

Like this: http://www.box.net/shared/9xuzp9tym7

Make a nice submenu structure following exactly the category structure of the Gnome Control Center. Like this: http://www.box.net/shared/9xuzp9tym7

Written by cousteau the 12 Feb 09 at 20:14.

Instead of just "Preferences" and "Administration", add more submenus (for example: "Preferences", "Hardware", "Look and feel", "Administration"). Like Solution #1 but using submenus instead of sub-submenus.

Instead of just "Preferences" and "Administration", add more submenus (for example: "Preferences", "Hardware", "Look and feel", "Administration"). Like Solution #1 but using submenus instead of sub-submenus.

Written by JimmyVolatile the 10 Feb 09 at 22:53.

It's by far the most newbie-friendly compared to the existing menu.

You can even type in search words to find what you're after, for instance "mouse" or "resolution".

Furthermore, it can easily be expanded with new items without risking an even more crowded menu.

It's by far the most newbie-friendly compared to the existing menu. You can even type in search words to find what you're after, for instance "mouse" or "resolution". Furthermore, it can easily be expanded with new items without risking an even more crowded menu.

Written by Endolith the 9 Mar 09 at 14:35.

Like solution #6, except without sub-sub-menus. Just put "Personal", "Look and feel", etc. directly under the System menu.

"Use submenus sparingly, as they are physically difficult to navigate and make it harder to find and reach the items they contain." "Do not nest submenus within submenus. More than two levels of hierarchy are difficult to memorize and navigate."

http://library.gnome.org/devel/hig-book/stable/menus-types.html.en#menu-type-su bmenu

Like solution #6, except without sub-sub-menus. Just put "Personal", "Look and feel", etc. directly under the System menu. "Use submenus sparingly, as they are physically difficult to navigate and make it harder to find and reach the items they contain." "Do not nest submenus within submenus. More than two levels of hierarchy are difficult to memorize and navigate." http://library.gnome.org/devel/hig-book/stable/menus-types.html.en#menu-type-submenu

Written by wish the 5 Mar 09 at 19:25.

1. For example, the Network configurations, Network Proxy and Remote Desktop can be grouped under a single listing Network.
2. Similarly, Compiz Settings Manager, Appearance, Emerald, Screen Saver, Screen Resolution, Windows etc can be grouped under GUI or some other better name.
3. Bluetooth, Printer, Mouse, Palm, etc can be grouped under devices.

Preferences and Administration should also get the option to be open in separate windows where they will not be cluttered.

1. For example, the Network configurations, Network Proxy and Remote Desktop can be grouped under a single listing Network. 2. Similarly, Compiz Settings Manager, Appearance, Emerald, Screen Saver, Screen Resolution, Windows etc can be grouped under GUI or some other better name. 3. Bluetooth, Printer, Mouse, Palm, etc can be grouped under devices. Preferences and Administration should also get the option to be open in separate windows where they will not be cluttered.

Written by amrlima the 2 Mar 09 at 20:30.

Fedora has a very elegant solution for this. They divide the menu in categories such as "Look and Feel" etc.

It really keeps the menu clean and it's easier to find what you are looking for.

Fedora has a very elegant solution for this. They divide the menu in categories such as "Look and Feel" etc. It really keeps the menu clean and it's easier to find what you are looking for.

Written by razor7 the 3 Mar 09 at 15:04.

Use the Gnome Control Center. The problem is that most users (like me) not even know that this cool Control Center even exists...Make it more visible and improve it

Use the Gnome Control Center. The problem is that most users (like me) not even know that this cool Control Center even exists...Make it more visible and improve it

Written by Xero Xenith the 3 Mar 09 at 16:08.

http://brainstorm.ubuntu.com/idea/80/

http://brainstorm.ubuntu.com/idea/80/

Written by probono the 4 Mar 09 at 22:09.

The separation between system-wide and per-user settings is not necessary. Neither Windows nor Mac OS X have this kind of separation, as it confuses the user more than it helps. I've observed many people to constantly search for a setting in the wrong menu, because nobody remembers in which menu a certain setting is. Hence, I propose to unify all settings in just one menu, as demonstrated in http://www.youtube.com/watch?v=AF2-JHF6v10

Also, the number of menu items should be brought down by combining multiple of these that fall under the same topic, as it is done in Mac OS X. For example, "Keyboard" and "Keyboard Shortcuts" could be tabs of the same panel. The same is true for "Software Sources" and "Synaptic Pagage Manger", to name just two examples.

The separation between system-wide and per-user settings is not necessary. Neither Windows nor Mac OS X have this kind of separation, as it confuses the user more than it helps. I've observed many people to constantly search for a setting in the wrong menu, because nobody remembers in which menu a certain setting is. Hence, I propose to unify all settings in just one menu, as demonstrated in http://www.youtube.com/watch?v=AF2-JHF6v10 Also, the number of menu items should be brought down by combining multiple of these that fall under the same topic, as it is done in Mac OS X. For example, "Keyboard" and "Keyboard Shortcuts" could be tabs of the same panel. The same is true for "Software Sources" and "Synaptic Pagage Manger", to name just two examples.

Written by James_Lochhead the 1 Apr 09 at 21:00.

Group similar entries into sub-menus. On average it would be easier to find things, the menus would look less cluttered and the menus would display well on lower resolutions.

I recently had a go at Fedora 10 (went back to Ubuntu within 10 minutes!) and the menus were implemented this way on Fedora.

Group similar entries into sub-menus. On average it would be easier to find things, the menus would look less cluttered and the menus would display well on lower resolutions. I recently had a go at Fedora 10 (went back to Ubuntu within 10 minutes!) and the menus were implemented this way on Fedora.

Written by hariks0 the 7 Apr 09 at 18:37.

If the menu preference could be set in a way that only the recent or frequent items are shown and the others only after a preset delay. Also the "Edit Main Menu" could be used to eliminate the unwanted items from popping up [by the user].

There should be a screen during installation letting the user select which applets they want to install with some tool tip for each of the items.Another requirement is an option to replace the default "menu Bar" with "Main Menu Icon" in initial setup from live CD ,which makes the desktop cleaner.

My third suggestion is to open the two menu items of"System" tabbed in to a window named "Preferences" the other "Administration". The applets/ items could be represented by icons and in alphabetical order.

If the menu preference could be set in a way that only the recent or frequent items are shown and the others only after a preset delay. Also the "Edit Main Menu" could be used to eliminate the unwanted items from popping up [by the user]. There should be a screen during installation letting the user select which applets they want to install with some tool tip for each of the items.Another requirement is an option to replace the default "menu Bar" with "Main Menu Icon" in initial setup from live CD ,which makes the desktop cleaner. My third suggestion is to open the two menu items of"System" tabbed in to a window named "Preferences" the other "Administration". The applets/ items could be represented by icons and in alphabetical order.

Written by nq6 the 24 May 09 at 21:35.

Below I developed some screens to explain better my idea.

Reduce the number of options available to facilitate the choice by the user. Decrease from 39 to 16 items.


Scheme simplification.


idea of how many icons can become one. view of the behavior of windows with the merger of these icons.


overview of the idea. In this picture you can see the whole blueprint at once.

Below I developed some screens to explain better my idea. Reduce the number of options available to facilitate the choice by the user. Decrease from 39 to 16 items. <a href="http://img30.imageshack.us/img30/6915/ideanq6.jpg" title="Nq6 idea"><img src="http://farm4.static.flickr.com/3324/3599989553_82e7e0512f.jpg" width="500" height="221" alt="idea nq6" /></a> Scheme simplification. <a href="http://img193.imageshack.us/img193/5403/96435707.jpg" title="Simplify the system of preferences"><img src="http://farm4.static.flickr.com/3648/3560259065_a876cc1324.jpg" width="500" height="267" alt="01" /></a> idea of how many icons can become one. view of the behavior of windows with the merger of these icons. <a href="http://img7.imageshack.us/img7/1473/77578921.jpg" title="Nq6 Idea"><img src="http://farm3.static.flickr.com/2467/3561072856_1ab7feeed0.jpg" width="500" height="206" alt="02" /></a> overview of the idea. In this picture you can see the whole blueprint at once. <a href="http://img268.imageshack.us/img268/4152/03a.jpg" title="Nq6 Idea"><img src="http://farm4.static.flickr.com/3602/3560255681_796d7a1961.jpg" width="500" height="474" alt="03a" /></a> Simplify the menu to facilitate usability. Below the simulation of how the menu of Ubuntu can be. Much simpler, less polluted Visually, and easier to find what you want. <a href="http://img268.imageshack.us/img268/4130/45298988.jpg" title="Nq6 Idea"><img src="http://farm4.static.flickr.com/3388/3561762040_de223eb9e1.jpg" width="341" height="500" alt="04" /></a> The Ubuntu is an excellent system, more needs to move forward in some areas. One is the simplification of the options that the user faces. The current options make Ubuntu confused. I think that my idea is to improve usability. <a href="http://img41.imageshack.us/img41/8748/ideanq62.jpg" title="Nq6 idea"><img src="http://farm4.static.flickr.com/3603/3599989715_5165cd6a29.jpg" width="500" height="317" alt="idea nq6 2" /></a>

Written by bryonak the 25 May 09 at 16:43.

Do as solution #17 proposes, except for the sub menus that pop up in a new window, which I find looks a bit awkward.

Instead of a popup, make the sub menus "slide out" below the clicked entry in the same window. The other entries just move further down.

Do as solution #17 proposes, except for the sub menus that pop up in a new window, which I find looks a bit awkward. Instead of a popup, make the sub menus "slide out" below the clicked entry in the same window. The other entries just move further down. <img src="http://img441.imageshack.us/img441/2034/slidemenus.jpg"></img>

Written by Frantique the 23 Jun 09 at 06:47.

As mentioned in the title, add a new tab, let's say after Visual Effects... Handy and easy to find the way of changing resolution of your screen.

As mentioned in the title, add a new tab, let's say after Visual Effects... Handy and easy to find the way of changing resolution of your screen.

Written by returnofnights the 23 Jun 09 at 08:55.

The tab should be there, new to Linux users expect to find screen controls in with the theme, effects and display properties. Changing the Context Menu item name to reflect the features: (Instead of 'Change Background' make it 'Display Settings') This is simple and to the point for new and old Gnome users.

The tab should be there, new to Linux users expect to find screen controls in with the theme, effects and display properties. Changing the Context Menu item name to reflect the features: (Instead of 'Change Background' make it 'Display Settings') This is simple and to the point for new and old Gnome users.

Written by je1403 the 14 Jul 09 at 23:41.

Linux Mint is an Ubuntu-based distribution. It is especially beginner-friendly. One of its best features is the mintMenu. This replaces the gnome-menu, but is categorized in the same way. It's pretty difficult to explain it by text, but check out this image: http://www.jessture.com/images/linux_mint/lm3-1_mint_menu.png
The menu is divided in several sections. The "Applications" section is categorized with the average "Office", "Internet", etc. Instead of having the pop-up menus, the menu has a fixed width and height, using scrollbars when the listed items dont fit. Mouseovering the categories shows their contents. The best part is the search bar. If you type "screen" it will filter the results and show you screen resolution, screensaver, login screen, for example. Watch this video:http://www.youtube.com/watch?v=ePEbfsV2yQw

Linux Mint is an Ubuntu-based distribution. It is especially beginner-friendly. One of its best features is the mintMenu. This replaces the gnome-menu, but is categorized in the same way. It's pretty difficult to explain it by text, but check out this image: http://www.jessture.com/images/linux_mint/lm3-1_mint_menu.png The menu is divided in several sections. The "Applications" section is categorized with the average "Office", "Internet", etc. Instead of having the pop-up menus, the menu has a fixed width and height, using scrollbars when the listed items dont fit. Mouseovering the categories shows their contents. The best part is the search bar. If you type "screen" it will filter the results and show you screen resolution, screensaver, login screen, for example. Watch this video:http://www.youtube.com/watch?v=ePEbfsV2yQw

Written by Aielyn the 26 Oct 10 at 13:15.

Instead of the current split into Preferences and Administration, have the System menu split into parts of the system. And then, on each submenu, the list would be split so that Preferences go on the top of the submenu, and Administration goes on the bottom, with a divider between them, thereby making it easier to distinguish between them, while being able to see both.

EXAMPLE LAYOUT - NOT PART OF SOLUTION ITSELF, JUST A POSSIBLE LAYOUT:

- Software
- Interface
- User
- System
- Third Party

Software would contain everything to do with software - Update Manager, Synaptic Package Manager, Software Sources, Computer Janitor, and Preferred Applications would all be examples.

Interface would contain everything to do with the User Interface - Mouse, Keyboard, Monitors, Audio, Menus, Appearance, etc.

User would hold things that are relevant to the users themselves - Languages, Passwords, About Me, E-Mail Settings, and Users and Groups as a few examples.

Instead of the current split into Preferences and Administration, have the System menu split into parts of the system. And then, on each submenu, the list would be split so that Preferences go on the top of the submenu, and Administration goes on the bottom, with a divider between them, thereby making it easier to distinguish between them, while being able to see both. EXAMPLE LAYOUT - NOT PART OF SOLUTION ITSELF, JUST A POSSIBLE LAYOUT: - Software - Interface - User - System - Third Party Software would contain everything to do with software - Update Manager, Synaptic Package Manager, Software Sources, Computer Janitor, and Preferred Applications would all be examples. Interface would contain everything to do with the User Interface - Mouse, Keyboard, Monitors, Audio, Menus, Appearance, etc. User would hold things that are relevant to the users themselves - Languages, Passwords, About Me, E-Mail Settings, and Users and Groups as a few examples. System would contain everything to do with system settings and underlying system operation - Time and Date, Drivers, power management, Startup Applications, System Monitor, etc. Third Party would then contain anything introduced by third-party programs and dealing specifically with those third-party programs - for instance, Opera Widget Manager, Wine settings, etc. TO BE CLEAR - THE ABOVE IS JUST AN EXAMPLE, NOT PART OF THE SOLUTION ITSELF. It can serve as a discussion point regarding how to best implement the Solution, but the Solution itself is only concerned with splitting it into categories and then dividing the resulting submenus into Preferences and Administration segments.

Written by Darwin Survivor the 7 Oct 10 at 17:10.

Change
-System
---Preferences
---Administration

to

-Settings
---User Settings
---System Settings

Change -System ---Preferences ---Administration to -Settings ---User Settings ---System Settings

Written by komputes the 8 Mar 11 at 18:19.

Broadcast (gwibber) and Messaging (empathy) preferences are accessible from within the respective applications. Remove preferences for these applications from the "System" menu.

Broadcast (gwibber) and Messaging (empathy) preferences are accessible from within the respective applications. Remove preferences for these applications from the "System" menu.