Written by aysiu the 28 Feb 08 at 18:51.

Ubuntu Brainstorm was updated in January 2009. Since the idea #141 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 #141 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 Ubuwu the 5 Feb 09 at 17:24.

StartUp Manager, or SUM, is a gui tool for changing settings for Grub, Grub2, Usplash and Splashy.

http://web.telia.com/~u88005282/sum/

StartUp Manager, or SUM, is a gui tool for changing settings for Grub, Grub2, Usplash and Splashy. http://web.telia.com/~u88005282/sum/

Written by spyyder the 16 Mar 08 at 18:27.

Ubuntu Brainstorm was updated in January 2009. Since the idea #4848 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 #4848 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 tloxscrew the 29 Feb 08 at 00:54.

Ubuntu Brainstorm was updated in January 2009. Since the idea #356 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 #356 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 cheesehead the 8 Apr 09 at 03:22.

The 'ionice' command sets the io scheduling class and priority for a program. A user can easily downgrade the io priority of any process. ionice is included in the 'schedutils' package, already in the repositories.
It's powerful and solves a lot of disk-thrashing if used properly. Read the man page before using it!

The 'ionice' command sets the io scheduling class and priority for a program. A user can easily downgrade the io priority of any process. ionice is included in the 'schedutils' package, already in the repositories. It's powerful and solves a lot of disk-thrashing if used properly. Read the man page before using it!

Written by antiriad_ubuntu the 16 Jan 09 at 17:18.

identify kio slaves activity with same protocol + src + dst and append action to a queue. maybe it's not suitable for all protocols but on file copying would be nice.

identify kio slaves activity with same protocol + src + dst and append action to a queue. maybe it's not suitable for all protocols but on file copying would be nice.

Written by idzuna the 19 Mar 09 at 23:18.

I suggest that there be a queue for copying between the same HDD's or even one's that are busy.

Example:
Task 1: HDD1 > HDD2 10 mins of copy time
Then the user wants to copy something from HDD1 > HDD3
Task 2: HDD1 > HDD3 5 mins of copy time

If they were in a queue (perhaps a queue that the user could edit in order of importance) the overall copy time wouldn't be affected.

also, if both.
HDD1 copies to HDD2
HDD3 copies to HDD4
then they could run simultaneously as the head of the HDD's is dedicated to each task.
Note: I say copy, but this can be applied to any HDD task

I suggest that there be a queue for copying between the same HDD's or even one's that are busy. Example: Task 1: HDD1 > HDD2 10 mins of copy time Then the user wants to copy something from HDD1 > HDD3 Task 2: HDD1 > HDD3 5 mins of copy time If they were in a queue (perhaps a queue that the user could edit in order of importance) the overall copy time wouldn't be affected. also, if both. HDD1 copies to HDD2 HDD3 copies to HDD4 then they could run simultaneously as the head of the HDD's is dedicated to each task. Note: I say copy, but this can be applied to any HDD task

Written by xubaj the 27 Aug 09 at 00:01.

clicking the pause button pauses the transfer
click-dragging down reduces the transfer speed
the pause button should indicate this feature with a little slider on the left side

reducing the speed is useful if the file transfer consumes to much CPU or HD time and you can't keep on working

clicking the pause button pauses the transfer click-dragging down reduces the transfer speed the pause button should indicate this feature with a little slider on the left side reducing the speed is useful if the file transfer consumes to much CPU or HD time and you can't keep on working

Written by xubaj the 27 Aug 09 at 00:08.

sometimes you start a transfer of some big files and while still transferring, you want to copy some small documents to work with. the small files would be enqueued and you'd have to wait for the big files to complete. in this case the small files should be prioritized.

sometimes you start a transfer of some big files and while still transferring, you want to copy some small documents to work with. the small files would be enqueued and you'd have to wait for the big files to complete. in this case the small files should be prioritized.

Written by cudjoe the 28 Nov 09 at 11:04.

I read somewhere that the nautilus code would not allow easily to manage the queue, because it relies on gvfs to run the move/copy process.

I propose that nautilus should offer a hook to define an gconf entry that specifies the move/copy utility to use.

Because there are several tools already, but they don't replace the default copy/move operation.
Eg. Minicopier, Ultracopier, ...

I read somewhere that the nautilus code would not allow easily to manage the queue, because it relies on gvfs to run the move/copy process. I propose that nautilus should offer a hook to define an gconf entry that specifies the move/copy utility to use. Because there are several tools already, but they don't replace the default copy/move operation. Eg. Minicopier, Ultracopier, ...

Written by XDarkness81 the 26 Apr 10 at 21:24.

Single click high priority (Move/copy this file/folder next )

Allow simultaneous copy's on differed devices. (checkbox)
For all that matter compression, encryptions and so on should be able to be placed in that queue to...
Only on system idle checkbox (set copy/move/extract/compress ... as nice level 15 )
And let the app remember the choices you made. (idle, simcopy)
(Or better, put them in a configuration dialog) keep the file action dialog simple)

Ah and put a hide to background after 3 seconds/on large transfers to... (maybe pop back up after a large transfer) Since now it only on top or minimize, i hate that!

Thanks nice idea, it has been buggin me to for a long time..

[EDIT!]
In light of hiding the information dialog...
After the copy/move dialog has calculated its ETA if that should be longer than 1 minute disable the keepontop. if its almost finished (for all files in queue) (let's say the last 10 seconds or 30) pop it back up.. "so you know.."
[/EDIT]

Single click high priority (Move/copy this file/folder next ) Allow simultaneous copy's on differed devices. (checkbox) For all that matter compression, encryptions and so on should be able to be placed in that queue to... Only on system idle checkbox (set copy/move/extract/compress ... as nice level 15 ) And let the app remember the choices you made. (idle, simcopy) (Or better, put them in a configuration dialog) keep the file action dialog simple) Ah and put a hide to background after 3 seconds/on large transfers to... (maybe pop back up after a large transfer) Since now it only on top or minimize, i hate that! Thanks nice idea, it has been buggin me to for a long time.. [EDIT!] In light of hiding the information dialog... After the copy/move dialog has calculated its ETA if that should be longer than 1 minute disable the keepontop. if its almost finished (for all files in queue) (let's say the last 10 seconds or 30) pop it back up.. "so you know.." [/EDIT]

Written by Ypthor the 12 May 10 at 04:30.

In case I want to do somethingelse while waiting for the transfer to finish. This could be automatic (but overridable), with nautilus watching if there is user activity, and raising io priority only when the user is idle.

In case I want to do somethingelse while waiting for the transfer to finish. This could be automatic (but overridable), with nautilus watching if there is user activity, and raising io priority only when the user is idle.

Written by ways the 28 May 10 at 09:53.

I agree with lots of these solutions, I just wantet to add something.

When you have one or more transfers going (or waiting in queue), allow drag'n'drop on the queue to enqueue more files to the same destination. Example here:

http://tvboks.no-ip.org/upload/files/copy_more_files2.png

I agree with lots of these solutions, I just wantet to add something. When you have one or more transfers going (or waiting in queue), allow drag'n'drop on the queue to enqueue more files to the same destination. Example here: http://tvboks.no-ip.org/upload/files/copy_more_files2.png

Written by ghostry the 4 Jun 10 at 09:08.

In the duplication/migration document window cancels the left side adds the suspension button
This not only solves the duplication/migration/elimination priority problem, moreover may when necessary reduce hard disk's busy degree, is advantageous for us fast to complete other work.

In the duplication/migration document window cancels the left side adds the suspension button This not only solves the duplication/migration/elimination priority problem, moreover may when necessary reduce hard disk's busy degree, is advantageous for us fast to complete other work.

Written by XDarkness81 the 13 Jul 10 at 20:58.

Don't know if this feature is implemented bud the file transfer manager could also keep in mind the filesystem its copying to, if i remember correct the new ext4 filesystem can reserve a few blocks of diskspace that can eliminate fragmentation.

maybe optional as plugins to the manager (and not that we have to select them later on, just add them and load them into memory if they are usefull (if filesystem exists/target fs is...)

Don't know if this feature is implemented bud the file transfer manager could also keep in mind the filesystem its copying to, if i remember correct the new ext4 filesystem can reserve a few blocks of diskspace that can eliminate fragmentation. maybe optional as plugins to the manager (and not that we have to select them later on, just add them and load them into memory if they are usefull (if filesystem exists/target fs is...)

Written by supercheetah the 24 Oct 10 at 09:47.

This idea is mostly in response to the problem that responsiveness in Linux goes down as disk activity goes up, and that is definitely an issue, but putting in a queue to fix the problem at the heart doesn't really fix the problem. I/O (disk activity) heavy applications are still going to cause slowdowns. I just played around with ionice on a few applications I run that are pretty I/O heavy, and it went a long way in resolving this issue for me, so I have an alternate solution here.

The I/O priority of applications can be put into three categories:

Idle (3): an application will only get I/O when when the system is idle. I won't recommend this as a default, but there may be some applications/programs where this makes sense as its default I/O priority.

Best Effort (2): Just as its name implies, and it has sub-classes (0-7, zero being highest priority). Round robin is used for any applications in the same class.

Real Time (1): The application's I/O takes precedence over anything else. Definitely *not* recommended.

Most applications will have a default I/O best effort nice priority based on their cpu nice priority (cpu_nice+20/5), which means for most user applications, they will have a best effort nice priority of five. I think this is too low. The cpu nice priority doesn't need to change, but I think the default ionice priority should be best effort seven.

This idea is mostly in response to the problem that responsiveness in Linux goes down as disk activity goes up, and that is definitely an issue, but putting in a queue to fix the problem at the heart doesn't really fix the problem. I/O (disk activity) heavy applications are still going to cause slowdowns. I just played around with ionice on a few applications I run that are pretty I/O heavy, and it went a long way in resolving this issue for me, so I have an alternate solution here. The I/O priority of applications can be put into three categories: Idle (3): an application will only get I/O when when the system is idle. I won't recommend this as a default, but there may be some applications/programs where this makes sense as its default I/O priority. Best Effort (2): Just as its name implies, and it has sub-classes (0-7, zero being highest priority). Round robin is used for any applications in the same class. Real Time (1): The application's I/O takes precedence over anything else. Definitely *not* recommended. Most applications will have a default I/O best effort nice priority based on their cpu nice priority (cpu_nice+20/5), which means for most user applications, they will have a best effort nice priority of five. I think this is too low. The cpu nice priority doesn't need to change, but I think the default ionice priority should be best effort seven.

Written by snecci the 8 Apr 12 at 13:52.

What a bout a general queue? Where you can enqueue any operation you want, like extracting a file, deleting the original compressed file, moving a file, unmounting a device, etc.

What a bout a general queue? Where you can enqueue any operation you want, like extracting a file, deleting the original compressed file, moving a file, unmounting a device, etc.

Written by snecci the 8 Apr 12 at 13:52.

What a bout a general queue? Where you can enqueue any operation you want, like extracting a file, deleting the original compressed file, moving a file, unmounting a device, etc.

What a bout a general queue? Where you can enqueue any operation you want, like extracting a file, deleting the original compressed file, moving a file, unmounting a device, etc.

Written by lejeunerandy the 29 Feb 08 at 10:37.

Ubuntu Brainstorm was updated in January 2009. Since the idea #1230 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 #1230 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 DrG the 20 Nov 09 at 06:48.

See http://brainstorm.ubuntu.com/idea/22555/

Unlike the windows system restore application. This Solution (the first part, other than the 'merge' option) can work without error as per my experience. Not sure about the 'merge' option.

See http://brainstorm.ubuntu.com/idea/22555/ Unlike the windows system restore application. This Solution (the first part, other than the 'merge' option) can work without error as per my experience. Not sure about the 'merge' option.

Written by SPM the 2 Aug 09 at 11:33.

The syncronization would be achieved by pushing package selection to the server after synaptic or apt completes, and using rsync for the rest. The home directory, and /etc directory that are backed up should be stored in an encrypted form (encrypted using the local machine administrative user password which should not be stored on the Canonical or Amazon website. There should be some indication to the user of the machine that it is being backed up or synchronized to the Internet so that the system cannot be set up to surreptitiously capture data by a third party who would register on the website and use the service as a trojan without the knowledge of the desktop's Ubuntu administrative user. The encryption of personal data also ensures privacy of personal data from the corporate system administrator (which a lot of managers would want). If individual files and directories are required to be backed up unencrypted to guard against loss of the local machine's administrative password, the an additional unencrypted rsync backup can be specified for those specific files and directories, and/or for all files in the /etc directory and /home directory files that are readable by "others".

The Canonical service which would not involve storing huge amounts of data as it would only include configuration and package data would not cost much, and the information on package use and customer base would probably pay for itself. Hence it should be viable to offer it free to registered users. In addition, data files would be stored on Amazon's S3 cloud storage, and Amazon could perhaps pay Canonical for the users directed to its S3 cloud storage.

The syncronization would be achieved by pushing package selection to the server after synaptic or apt completes, and using rsync for the rest. The home directory, and /etc directory that are backed up should be stored in an encrypted form (encrypted using the local machine administrative user password which should not be stored on the Canonical or Amazon website. There should be some indication to the user of the machine that it is being backed up or synchronized to the Internet so that the system cannot be set up to surreptitiously capture data by a third party who would register on the website and use the service as a trojan without the knowledge of the desktop's Ubuntu administrative user. The encryption of personal data also ensures privacy of personal data from the corporate system administrator (which a lot of managers would want). If individual files and directories are required to be backed up unencrypted to guard against loss of the local machine's administrative password, the an additional unencrypted rsync backup can be specified for those specific files and directories, and/or for all files in the /etc directory and /home directory files that are readable by "others". The Canonical service which would not involve storing huge amounts of data as it would only include configuration and package data would not cost much, and the information on package use and customer base would probably pay for itself. Hence it should be viable to offer it free to registered users. In addition, data files would be stored on Amazon's S3 cloud storage, and Amazon could perhaps pay Canonical for the users directed to its S3 cloud storage.

Written by dr3mro the 12 Apr 10 at 01:13.

2 new items

1. undo last synaptic ( system restore )
2. restore Ubuntu from backup image

2 new items 1. undo last synaptic ( system restore ) 2. restore Ubuntu from backup image

Written by SeySayux the 29 Feb 08 at 10:54.

Other OS'es (Windows, Mac OS X, yes even SUSE, ...) have a "Repair existing system" option. This allows to reinstall the OS without having to reformat. All it would have to do is to manually check and repair dpkg/apt-get, next it has to reinstall every package present on the cd. Very easy, not too hard to implement, many users that would be very happy. Why not do it?

Other OS'es (Windows, Mac OS X, yes even SUSE, ...) have a "Repair existing system" option. This allows to reinstall the OS without having to reformat. All it would have to do is to manually check and repair dpkg/apt-get, next it has to reinstall every package present on the cd. Very easy, not too hard to implement, many users that would be very happy. Why not do it?

Written by math the 4 Jan 09 at 21:00.

And then I would like to have the ALT+tab function enabled while dragging, in order to drop the file into the new email window.

And then I would like to have the ALT+tab function enabled while dragging, in order to drop the file into the new email window.

Written by loonyphoenix the 18 Jan 09 at 07:32.

Make it possible to hover while drag-and-dropping over the desired task in the taskbar or over a window in the background to bring the application to the front; or over a viewport in viewport selector to switch to it. No need for a keyboard!

Make it possible to hover while drag-and-dropping over the desired task in the taskbar or over a window in the background to bring the application to the front; or over a viewport in viewport selector to switch to it. No need for a keyboard!

Written by jharris1993 the 24 Jan 09 at 22:00.

Both of these solutions have merit.
Being a (eek!) Windoze user, I am used to being able to hover over a window and drop there even if the window doesn't have focus.

What I would like to see:
(a) Allow Alt-Tab as above
(b) Allow Hover as above
(c) Allow "edge awareness" when dragging.
(i.e. If the place you want to drag to is not visible within the window, dragging to the edge where it's hidden should cause the window's viewport to scroll in that direction.)

I have noticed this a number of times - especially when trying to select text by dragging - that the selection process simply STOPS at the edge instead of scrolling the viewpoint.

What say ye?

Jim

Both of these solutions have merit. Being a (eek!) Windoze user, I am used to being able to hover over a window and drop there even if the window doesn't have focus. What I would like to see: (a) Allow Alt-Tab as above (b) Allow Hover as above (c) Allow "edge awareness" when dragging. (i.e. If the place you want to drag to is not visible within the window, dragging to the edge where it's hidden should cause the window's viewport to scroll in that direction.) I have noticed this a number of times - especially when trying to select text by dragging - that the selection process simply STOPS at the edge instead of scrolling the viewpoint. What say ye? Jim

Written by ziroday the 14 Jan 09 at 11:50.

Before upgrading a system have a dialog box warning the user about regressions and a large link to the release notes so that they can check for any identified regressions against their hardware. It should also contain information on how to view what hardware you currently have (possibly through hardinfo or something similar).

Before upgrading a system have a dialog box warning the user about regressions and a large link to the release notes so that they can check for any identified regressions against their hardware. It should also contain information on how to view what hardware you currently have (possibly through hardinfo or something similar).

Written by Seph_VII the 14 Jan 09 at 21:14.

Before upgrading or installing Ubuntu, make it check an online(or on-cd, if installing from a LiveCD) blacklist of incompatible hardware. If incompatible hardware is found, make Ubuntu warn the user, and ask whether he/she still wants to continue.

Before upgrading or installing Ubuntu, make it check an online(or on-cd, if installing from a LiveCD) blacklist of incompatible hardware. If incompatible hardware is found, make Ubuntu warn the user, and ask whether he/she still wants to continue.

Written by ruben the 26 Jan 09 at 21:09.

The function i have in mind is a simple undo of an update or even a package installation.
Unlike apt-get --perge remove it would also delete any unneaded dependancies simmilar to autoremove. However this would make it possible to install updates and then if it didn't work undo the change. Including any movement of files or changes in other files.

The problem i see with an upgrade advisor is that it can never actually say if it will work as only trial and error can. Or at least in most cases. Also it is very possible that the upgrade advisor does not have all the correct information for all systems and thus advises incorrectly. Furthermore advice given need to be based on information gather beforehand. Thus an easy undo feature would make upgrading a lot less risky.

It would be even better if this feature could some how be accessed from recovery mode or a live cd to repair if the system was rendered unboot able. This feature should be a used in conjunction with an upgrade advisor. Perhaps more as a long run solution

The function i have in mind is a simple undo of an update or even a package installation. Unlike apt-get --perge remove it would also delete any unneaded dependancies simmilar to autoremove. However this would make it possible to install updates and then if it didn't work undo the change. Including any movement of files or changes in other files. The problem i see with an upgrade advisor is that it can never actually say if it will work as only trial and error can. Or at least in most cases. Also it is very possible that the upgrade advisor does not have all the correct information for all systems and thus advises incorrectly. Furthermore advice given need to be based on information gather beforehand. Thus an easy undo feature would make upgrading a lot less risky. It would be even better if this feature could some how be accessed from recovery mode or a live cd to repair if the system was rendered unboot able. This feature should be a used in conjunction with an upgrade advisor. Perhaps more as a long run solution

Written by torkiano the 30 Jan 09 at 20:45.

Smolt is a hardware profiler to enable users to submit their hardware profiles during installation.

Smolt, like PackageKit from Fedora is also a distribution neutral tool and collects stats anonymously and sends it to a central database.

It became clear quite quickly that it does not make sense to have a per-distro solution for that - if we want to have momentum with a hardware database a combined effort promises the most.

Fedora and Opensuse already implemented it.

See http://smolts.org/
http://www.osnews.com/story/20621/Smolt_gets_adopted_by_openSUSE

Smolt is a hardware profiler to enable users to submit their hardware profiles during installation. Smolt, like PackageKit from Fedora is also a distribution neutral tool and collects stats anonymously and sends it to a central database. It became clear quite quickly that it does not make sense to have a per-distro solution for that - if we want to have momentum with a hardware database a combined effort promises the most. Fedora and Opensuse already implemented it. See http://smolts.org/ http://www.osnews.com/story/20621/Smolt_gets_adopted_by_openSUSE

Written by pi314 the 29 Feb 08 at 13:28.

Ubuntu Brainstorm was updated in January 2009. Since the idea #1388 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 #1388 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 inquata the 2 Jun 10 at 07:41.

While this is not implemented, the Nautilus script »Renamer« is a nice workaround to have something for batch renaming directly in Nautilus:
http://gnome-look.org/content/show.php/Renamer?content=87601

Sadly, that still does not make batch renaming as easy as in Windows.

While this is not implemented, the Nautilus script »Renamer« is a nice workaround to have something for batch renaming directly in Nautilus: http://gnome-look.org/content/show.php/Renamer?content=87601 Sadly, that still does not make batch renaming as easy as in Windows.

Written by Lain_13 the 31 Jul 10 at 06:52.

There was idea in comments about it. So, here is solution to use it. This tool as powerful as GPrename but simpler and cleaner. In thunar you need to select few files and press F2 (rename from menu) to invoke it.

There was idea in comments about it. So, here is solution to use it. This tool as powerful as GPrename but simpler and cleaner. In thunar you need to select few files and press F2 (rename from menu) to invoke it.

Written by zaryk the 29 Feb 08 at 15:57.

Ubuntu Brainstorm was updated in January 2009. Since the idea #1521 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 #1521 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 xubaj the 27 Aug 09 at 01:18.

i don't know if Linux does this already but applications that are used frequently like Nautilus should be in RAM all the time.

the folder contents should be cached. the content may be changed by other applications (terminal) and Nautilus will loose track of the current content. either Nautilus should keep track of folder content even when closed or just fake the content based on the previous cache and already allow user interaction while streaming the current content in background.

an example:
content of /tmp is 100 files
Nautilus gets closed
in terminal i add 50 files
reopen Nautilus and navigate to /tmp
show old cache of 100 files and stream current state in background
if i click another folder (/tmp/test) or file while streaming, the current streaming is stopped and the new folder opened and thus wasting no time with folder updating when i already know where i want to go (/tmp/test)

i don't know if Linux does this already but applications that are used frequently like Nautilus should be in RAM all the time. the folder contents should be cached. the content may be changed by other applications (terminal) and Nautilus will loose track of the current content. either Nautilus should keep track of folder content even when closed or just fake the content based on the previous cache and already allow user interaction while streaming the current content in background. an example: content of /tmp is 100 files Nautilus gets closed in terminal i add 50 files reopen Nautilus and navigate to /tmp show old cache of 100 files and stream current state in background if i click another folder (/tmp/test) or file while streaming, the current streaming is stopped and the new folder opened and thus wasting no time with folder updating when i already know where i want to go (/tmp/test)

Written by Zatara214 the 9 Sep 09 at 23:14.

As per my comment below, I think OSX does something similar. To take it even further, maybe AWN could develop integration for this if implemented. This would be the optimal way of using this idea while still appealing to both sides of the argument.

As per my comment below, I think OSX does something similar. To take it even further, maybe AWN could develop integration for this if implemented. This would be the optimal way of using this idea while still appealing to both sides of the argument.

Written by Zatara214 the 10 Sep 09 at 05:56.

Come on, Firefox loads REALLY slowly and we all know it. Even with Preload installed, it's definitely one of the slowest loading programs installed by default (along with OpenOffice). If it was kept in memory, I think many more people would be willing to switch to Ubuntu based solely on the speed of the browser. Again, this should be optional, as it wouldn't help on older machines or machines with limited RAM.

Come on, Firefox loads REALLY slowly and we all know it. Even with Preload installed, it's definitely one of the slowest loading programs installed by default (along with OpenOffice). If it was kept in memory, I think many more people would be willing to switch to Ubuntu based solely on the speed of the browser. Again, this should be optional, as it wouldn't help on older machines or machines with limited RAM.

Written by adisk the 16 Sep 09 at 16:09.

Add option for disable MIME detection.
File type detection is very slow? because need open and read most files.
Detect MIME on click only. (on left-click and on right-click)

Option like "Disable MIME detection (for slow PC)"

Add option for disable MIME detection. File type detection is very slow? because need open and read most files. Detect MIME on click only. (on left-click and on right-click) Option like "Disable MIME detection (for slow PC)"

Written by Lachu the 28 Sep 09 at 13:40.

Disable mime detection for some system directories, like /bin, /sbin, etc.

Disable mime detection for some system directories, like /bin, /sbin, etc.

Written by Lachu the 8 Nov 09 at 19:59.

This will lack many think, because application can modify modification date, but also speed up folder opening operation.

Create cache and use kernel buffers. When application using GNOME file system layer, it's a great opportunity to regenerate preview, when file is closed and been modified. It can works for small files. How? Application closes file, using GNOME calls. GNOME will check if file was changed and regenerate preview, storing last modification date in special place.

When we open a folder, Nautilus compare only modification date. If user using only GNOME, all preview will been generated.

This will lack many think, because application can modify modification date, but also speed up folder opening operation. Create cache and use kernel buffers. When application using GNOME file system layer, it's a great opportunity to regenerate preview, when file is closed and been modified. It can works for small files. How? Application closes file, using GNOME calls. GNOME will check if file was changed and regenerate preview, storing last modification date in special place. When we open a folder, Nautilus compare only modification date. If user using only GNOME, all preview will been generated.

Written by mikko.rantalainen the 7 Oct 10 at 07:20.

The problem is very similar to the problem of rendering a web page: the process can get the rough content (a HTML page) quickly, but the referred content takes more time. Implement the same fix for the problem ("insert and render linked images / other resources as they come"):

Incremental rendering: immediately after the nautilus has received a file name from the OS, it should start rendering it. If current view mode is "Icons", the icon could first appear as empty/fully transparent rectangle of suitable size and later replaced with the real icon once the system knows it. The current implementation seems to be "never render anything until the full display with all the icons and stuff can be rendered".

Additional notes: try to make sure that label/icon placement is static from the start (reserve enough space for the final icon from the start), it would make sense to wait for 0.25 seconds before starting incremental rendering (compare to FOUC in web pages), improve caching (if nautilus has rendered /usr/bin 10 seconds ago, it shouldn't start the process from the start to get the correct icons and metadata), and finally animate (fade) between the different rendering states to make appearing icons less irritating.

The problem is very similar to the problem of rendering a web page: the process can get the rough content (a HTML page) quickly, but the referred content takes more time. Implement the same fix for the problem ("insert and render linked images / other resources as they come"): Incremental rendering: immediately after the nautilus has received a file name from the OS, it should start rendering it. If current view mode is "Icons", the icon could first appear as empty/fully transparent rectangle of suitable size and later replaced with the real icon once the system knows it. The current implementation seems to be "never render anything until the full display with all the icons and stuff can be rendered". Additional notes: try to make sure that label/icon placement is static from the start (reserve enough space for the final icon from the start), it would make sense to wait for 0.25 seconds before starting incremental rendering (compare to FOUC in web pages), improve caching (if nautilus has rendered /usr/bin 10 seconds ago, it shouldn't start the process from the start to get the correct icons and metadata), and finally animate (fade) between the different rendering states to make appearing icons less irritating.

Written by mikko.rantalainen the 7 Oct 10 at 07:31.

Currently opening a folder full of high resolution images take ages to get the thumbnails. Thumbnailers should be improved to support generating thumbnails with different priorities. Actually, two modes should be enough: fast and high quality.

Fast mode: get a thumbnail that somehow represents the real content (e.g. get the thumbnail from exif data or get the first pixel from each 8x8 pixel macroblock without computing the iDCT at all in case of JPEG, text only rendering of an HTML file) and generate thumbnail from that.

High quality mode: use as much CPU power as required to get the finest thumbnail that one can get. For example, a JPEG should be fully decoded, scaled with Sinc or other high quality scaler and rendered as final thumbnail. An M2TS file could be rendered as a small animation created from high quality thumbnails of 10 evenly separated screenshots from the file. A HTML file could be rendered by a browser and a screenshot should be used as an icon. In case of files in /usr/bin, a high quality screenshot would be a screenshot of UI combined with actual icon used by the application (I'm not sure these can be automatically generated at all, though -- possibly running the application in a sand box or in a virtual machine and getting a screenshot from it after letting it to get idle?).

This basically builds on Solution #8 above: render the contents of a folder without images at first, then with fast thumbnails and then with high quality thumbnails as they get ready. As a result, any folder would appear practically instantly, displayed with low quality thumbnails rapidly and rendered with high quality thumbnails if user waits for long enough (or high quality thumbnails can be already found from the cache).

Currently opening a folder full of high resolution images take ages to get the thumbnails. Thumbnailers should be improved to support generating thumbnails with different priorities. Actually, two modes should be enough: fast and high quality. Fast mode: get a thumbnail that somehow represents the real content (e.g. get the thumbnail from exif data or get the first pixel from each 8x8 pixel macroblock without computing the iDCT at all in case of JPEG, text only rendering of an HTML file) and generate thumbnail from that. High quality mode: use as much CPU power as required to get the finest thumbnail that one can get. For example, a JPEG should be fully decoded, scaled with Sinc or other high quality scaler and rendered as final thumbnail. An M2TS file could be rendered as a small animation created from high quality thumbnails of 10 evenly separated screenshots from the file. A HTML file could be rendered by a browser and a screenshot should be used as an icon. In case of files in /usr/bin, a high quality screenshot would be a screenshot of UI combined with actual icon used by the application (I'm not sure these can be automatically generated at all, though -- possibly running the application in a sand box or in a virtual machine and getting a screenshot from it after letting it to get idle?). This basically builds on Solution #8 above: render the contents of a folder without images at first, then with fast thumbnails and then with high quality thumbnails as they get ready. As a result, any folder would appear practically instantly, displayed with low quality thumbnails rapidly and rendered with high quality thumbnails if user waits for long enough (or high quality thumbnails can be already found from the cache).

Written by strattonbrazil the 14 Jul 08 at 22:28.

Ubuntu Brainstorm was updated in January 2009. Since the idea #11133 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 #11133 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 jarko_ the 22 Feb 09 at 12:46.

Like in some other systems, associate key combination to bring up 'recovery screen' or similary named one.

This window doesn't have to be normal GTK-window, but something from the upper level, like GDM (or xorg if going for extreme) created 'recovery screen/task manager'. This should ensure that no window or full screen application could hide or block the 'recovery screen'.

This recovery screen could have options to kill programs, log out, shutdown and lock computer etc.

Like in some other systems, associate key combination to bring up 'recovery screen' or similary named one. This window doesn't have to be normal GTK-window, but something from the upper level, like GDM (or xorg if going for extreme) created 'recovery screen/task manager'. This should ensure that no window or full screen application could hide or block the 'recovery screen'. This recovery screen could have options to kill programs, log out, shutdown and lock computer etc.

Written by Magnes the 23 Feb 09 at 07:46.

If there is an application that uses all the resources allow closing it (if it uses all the memory) or make it nicer (if it uses all the CPU) by one click in System Monitor.

If there is an application that uses all the resources allow closing it (if it uses all the memory) or make it nicer (if it uses all the CPU) by one click in System Monitor.

Written by jamessnell the 26 Feb 09 at 20:37.

== Graphical ==
If a separate X session configured entirely to show a System Monitor could be very well insulated from buggy applications.

In those cases where a game causes you to change video modes and then screws up, you'll still have a graceful way to touch the System Monitor.

== Command-Line ==
An ncurses interface (like that of "aptitude") for the command line would also be awesome for those cases when the entire Xorg system ignores you. This would help when the keyboard is being ignored as it'd be fast to pull up over ssh.


Yes this is could basically be a wrapper interface to ps.

== Graphical == If a separate X session configured entirely to show a System Monitor could be very well insulated from buggy applications. In those cases where a game causes you to change video modes and then screws up, you'll still have a graceful way to touch the System Monitor. == Command-Line == An ncurses interface (like that of "aptitude") for the command line would also be awesome for those cases when the entire Xorg system ignores you. This would help when the keyboard is being ignored as it'd be fast to pull up over ssh. Yes this is could basically be a wrapper interface to ps.

Written by zooounds the 3 Mar 09 at 11:15.

It rest in a tty until needed and can be used to kill application even if X is totaly broken.

It rest in a tty until needed and can be used to kill application even if X is totaly broken.

Written by cousteau the 5 Mar 09 at 17:27.

Ctrl+Alt+Del opens the Logout dialog (at least on Hardy). So it would be nice to add an "Open System Monitor" option to it.

Ctrl+Alt+Del opens the Logout dialog (at least on Hardy). So it would be nice to add an "Open System Monitor" option to it.

Written by jackmcslay the 7 Mar 09 at 14:57.

This is a continuation of #2. We should get xkill and gnome-system-monitor among the available action options on
System>Preferences>Keyboard shortcuts

so, even if Ctrl+Alt+Del remains as "logout" shortcut, the option of binding it to Ctrl+Alt+Del remains available

This is a continuation of #2. We should get xkill and gnome-system-monitor among the available action options on System>Preferences>Keyboard shortcuts so, even if Ctrl+Alt+Del remains as "logout" shortcut, the option of binding it to Ctrl+Alt+Del remains available

Written by Tom Mann the 12 Mar 09 at 20:47.

In KDE and XFCE, if you hit CTRL-ALT-ESC, your cursor turns into an X (or a skull and crossbones) and clicking any app (it doesn't have to be stuck) kills it.

I still don't get why it disappeared from Ubuntu's Gnome Desktop (I'm not sure if this happens on any other distros Gnome desktop)

In KDE and XFCE, if you hit CTRL-ALT-ESC, your cursor turns into an X (or a skull and crossbones) and clicking any app (it doesn't have to be stuck) kills it. I still don't get why it disappeared from Ubuntu's Gnome Desktop (I'm not sure if this happens on any other distros Gnome desktop)

Written by quartz the 20 Mar 09 at 17:12.

The system monitor window needs to be responsive, not just there.

Just bringing up system monitor (or a new manager if necessary) is not enough is the CPU is totally taken, if it comes up, any process(es) that might be hogging resources should be reniced to a slightly lower priority and the system monitor process should be started fairly high.

(A good question is what to do if the problem is with X itself, since renicing it might slow down the system monitor too)

The system monitor window needs to be responsive, not just there. Just bringing up system monitor (or a new manager if necessary) is not enough is the CPU is totally taken, if it comes up, any process(es) that might be hogging resources should be reniced to a slightly lower priority and the system monitor process should be started fairly high. (A good question is what to do if the problem is with X itself, since renicing it might slow down the system monitor too)

Written by Craig73 the 21 Mar 09 at 14:28.

Capture CTRL+ALT+DEL (or perhaps the second CTRL+ALT+DEL for just "frozen systems") at the kernel level, which opens a graphical task manager (logoff / process manager / whatever) which draws directly to the screen (bypassing X which may be frozen)

To implement this - it would write to the framebuffer, and would likely require KMS and DRI2. [Ideally it would capture the current screen in the framebuffer, and draw the dialog on top, for a integrated feeling].

Then have it fall back to VGA text only if it can't grab a graphical framebuffer (things are really hurting)

My intent is to handle cases such as X being frozen, or in a full screen game, etc.

Capture CTRL+ALT+DEL (or perhaps the second CTRL+ALT+DEL for just "frozen systems") at the kernel level, which opens a graphical task manager (logoff / process manager / whatever) which draws directly to the screen (bypassing X which may be frozen) To implement this - it would write to the framebuffer, and would likely require KMS and DRI2. [Ideally it would capture the current screen in the framebuffer, and draw the dialog on top, for a integrated feeling]. Then have it fall back to VGA text only if it can't grab a graphical framebuffer (things are really hurting) My intent is to handle cases such as X being frozen, or in a full screen game, etc.

Written by qwerty800 the 11 Jun 09 at 21:17.

It would be really nice to have a shortcut like Alt+SysRq+X, that automatically kill the focused application. Using a such shortcut would avoid the inconvenience of passing trough the task manager, nor restart the whole X server and to work with the full screen programs! Having a 16:10 monitor often causes me to get stuck with unsupported resolutions. When that happens, I have to restart my whole X server and THAT'S annoying!

Plus, "X" is easy to remember, because:
*It's not currently used.
*It can refer to Xkill
*It can refer to Xorg
*It can refer to the Window decoration (X=Close)!

It would be really nice to have a shortcut like Alt+SysRq+X, that automatically kill the focused application. Using a such shortcut would avoid the inconvenience of passing trough the task manager, nor restart the whole X server and to work with the full screen programs! Having a 16:10 monitor often causes me to get stuck with unsupported resolutions. When that happens, I have to restart my whole X server and THAT'S annoying! Plus, "X" is easy to remember, because: *It's not currently used. *It can refer to Xkill *It can refer to Xorg *It can refer to the Window decoration (X=Close)!

Written by Shnatsel the 25 Jul 09 at 10:09.

Xfce did so, and if something hangs, it's easy to kill it. GNOME has a panel applet for such purposes, but if a fullscreen game hangs, it's useless.

Xfce did so, and if something hangs, it's easy to kill it. GNOME has a panel applet for such purposes, but if a fullscreen game hangs, it's useless.

Written by viva.amego the 5 Mar 10 at 03:37.

its good idea and there is an easy solution for now


right click (system > administration > system monitor)
select (add this launcher to panel)
right click (system monitor "in panel")
select (properties )
copy ( command )

run (system > preference > keyboard shortcuts)
click ( add)

name -> system monitor
command -> right click( paste)

click (apply)

and you had shourtcut

its good idea and there is an easy solution for now right click (system > administration > system monitor) select (add this launcher to panel) right click (system monitor "in panel") select (properties ) copy ( command ) run (system > preference > keyboard shortcuts) click ( add) name -> system monitor command -> right click( paste) click (apply) and you had shourtcut

Written by Lachu the 28 Mar 10 at 17:06.

Simply start a new X Session for every application using fullscreen mode.

Simply start a new X Session for every application using fullscreen mode.

Written by trezker the 28 Mar 10 at 05:57.

Simply put, whenever an app makes a call to set a fullscreen mode the system refuses to do it.

Simply put, whenever an app makes a call to set a fullscreen mode the system refuses to do it.

Written by dottornik the 27 Apr 10 at 14:51.

Gizmod intercepts ctrl-alt-del (even when fullscreen application have focus) and execute:
wmctrl -k on (minimize all windows)
gnome-system-monitor

tested in Lucid & Karmic and work with:
-XBMC fullscreen crashed by youtube plugin
-fullscreen crashed gmameui
-fullscreen crashed flash player
-lot of crashed stuff

Instructions:

-install wmctrl
sudo apt-get install wmctrl

-disable default ctrl-alt-del shortcut from System>Preferences>Keyboard shortcuts

-install gizmod
http://gizmod.sourceforge.net/
install and set group permissions (see HOWTO - Setting Input Device Permissions - Creating a udev Rule)

Gizmod intercepts ctrl-alt-del (even when fullscreen application have focus) and execute: wmctrl -k on (minimize all windows) gnome-system-monitor tested in Lucid & Karmic and work with: -XBMC fullscreen crashed by youtube plugin -fullscreen crashed gmameui -fullscreen crashed flash player -lot of crashed stuff Instructions: -install wmctrl sudo apt-get install wmctrl -disable default ctrl-alt-del shortcut from System>Preferences>Keyboard shortcuts -install gizmod http://gizmod.sourceforge.net/ install and set group permissions (see HOWTO - Setting Input Device Permissions - Creating a udev Rule) -edit 199-Keyboard-Default.py in ~/.gizmod/gizmod/modules.d/ add this: elif Gizmo.getKeyState(GizmoKey.KEY_DELETE) >= 1 and Gizmo.getKeyState(GizmoKey.KEY_LEFTCTRL) >= 1 and Gizmo.getKeyState(GizmoKey.KEY_LEFTALT) >= 1: subprocess.Popen(["wmctrl", "-k", "on"]) subprocess.Popen(["gnome-system-monitor"]) return True This is a python script, please respect document indentation! Excuse me for my bad english... coming soon: automatic install and config script....

Written by jez9999 the 28 Feb 08 at 23:50.

Me, I still maintain that the storage should just be mounted synchronously. I *want* to be able to just pull the thing out when its light stops flashing (and it's VERY likely to be in a safe state to remove), if for no other reason than that I know I'll forget to unmount occasionally and I don't want my drive's filesystem screwed (or data lost) because of a one-off mistake. I do not think it is going to 'damage' flash media significantly, yeah it will result in a little more wear but frankly, so what? People only use flash dongles for maybe a year and most modern ones will last enough write cycles that this extra wear won't be an issue.

Me, I still maintain that the storage should just be mounted synchronously. I *want* to be able to just pull the thing out when its light stops flashing (and it's VERY likely to be in a safe state to remove), if for no other reason than that I know I'll forget to unmount occasionally and I don't want my drive's filesystem screwed (or data lost) because of a one-off mistake. I do not think it is going to 'damage' flash media significantly, yeah it will result in a little more wear but frankly, so what? People only use flash dongles for maybe a year and most modern ones will last enough write cycles that this extra wear won't be an issue.

Written by Endolith the 8 Apr 09 at 15:30.

Apparently sync mode has several disadvantages, like reducing life of the drive and greatly reduced speed in some situations. Here's an alternative:

Any time a drive is removed without unmounting first, a big scary warning should pop up telling the user they may have lost data or screwed up their drive. (I think this is what OS X does.) It should point out clearly, with an animation, how to unmount the drive before physically removing it, so they don't have any problems in the future.

The first time a USB drive is plugged in and auto-mounted, a notification should be displayed along with the contents of the drive, explaining the same issue, with a "do not show this anymore" button, to educate the user and avoid causing any problems even the first time.

Unmounting the drive should be made more obvious and intuitive. A right-click context menu is not obvious.

When a drive has been plugged in that hasn't been synced and is screwed up, tools should be immediately available to fix the problem, and an explanation should appear that points the user to where they can get help.

The word "unmount" should be changed to something more intuitive ("detach", "disconnect", "turn off").

Apparently sync mode has several disadvantages, like reducing life of the drive and greatly reduced speed in some situations. Here's an alternative: Any time a drive is removed without unmounting first, a big scary warning should pop up telling the user they may have lost data or screwed up their drive. (I think this is what OS X does.) It should point out clearly, with an animation, how to unmount the drive before physically removing it, so they don't have any problems in the future. The first time a USB drive is plugged in and auto-mounted, a notification should be displayed along with the contents of the drive, explaining the same issue, with a "do not show this anymore" button, to educate the user and avoid causing any problems even the first time. Unmounting the drive should be made more obvious and intuitive. A right-click context menu is not obvious. When a drive has been plugged in that hasn't been synced and is screwed up, tools should be immediately available to fix the problem, and an explanation should appear that points the user to where they can get help. The word "unmount" should be changed to something more intuitive ("detach", "disconnect", "turn off").

Written by pato101 the 3 Jun 09 at 12:24.

Not sure if this is feasible, and for sure requires kernel hacking and solving lots of small issues, but I think it is the best solution:

1. The device would be mounted async.
2. Kernel would keep track of status if device is unplugged.
3. In such case a message would ask the user to plug the device again only if there are pending operations
4. Then, kernel would see that the device has been plugged again and would end the operation.

Not sure if this is feasible, and for sure requires kernel hacking and solving lots of small issues, but I think it is the best solution: 1. The device would be mounted async. 2. Kernel would keep track of status if device is unplugged. 3. In such case a message would ask the user to plug the device again only if there are pending operations 4. Then, kernel would see that the device has been plugged again and would end the operation.

Written by slashdotaccount the 2 Aug 09 at 10:18.

Just lost the data just like ejecting the floppy disk do. Why not?

Just lost the data just like ejecting the floppy disk do. Why not?

Written by md5encrypted the 19 Jul 10 at 16:12.

And mount removable drives with `flush` option. Copying to drive will result in indicator growth which then return to 0 as data flushed. When it reaches 0, desktop environment should inform user with something like "it is now safe to remove drive" (and maybe change color of indicator icon)

And mount removable drives with `flush` option. Copying to drive will result in indicator growth which then return to 0 as data flushed. When it reaches 0, desktop environment should inform user with something like "it is now safe to remove drive" (and maybe change color of indicator icon)

Written by mikko.rantalainen the 7 Oct 10 at 07:58.

The synchronous mounting would be required only because kernel is too dumb to not lose the data. Fix this problem and it really does not make any sense to mount removable devices synchronously. To the contrary - mounting a flash device synchronously will decrease it's lifespan because of increased amount of writes.

Very similar to solution #3: See idea http://brainstorm.ubuntu.com/idea/1515/ Allow repairing incorrectly removed USB Mass Storage Device (flash memory stick.

The synchronous mounting would be required only because kernel is too dumb to not lose the data. Fix this problem and it really does not make any sense to mount removable devices synchronously. To the contrary - mounting a flash device synchronously will decrease it's lifespan because of increased amount of writes. Very similar to solution #3: See idea <a href="http://brainstorm.ubuntu.com/idea/1515/">http://brainstorm.ubuntu.com/idea/1515/ Allow repairing incorrectly removed USB Mass Storage Device (flash memory stick</a>.