Contributor Alberto Ferreira
"Live CD Size as an end" is harming ubuntu's image
Written by vexorian the 27 Jun 09 at 13:48.
Global category: Installation.
Implemented
Take a look at:
https://wiki.ubuntu.com/DesktopTeam/Meeting/2009-06-09#The%20GIMP
It is not the first time that a completely ridiculous, non-sense, anti-usability and "coincidentally" pro-Mono proposal (also coincidentally this time, from a former Microsoft employee) is pushed under the excuse of live cd size.
Apparently live CD size has become such an end instead of a means that ubuntu even considered removing the whole image editing thing to replace it with a sub-par photo management thingy. (No, apparently creation of images is too complicated for ubuntu users, they only want to edit a couple of pictures, let creativity be part of the windows or mac OS/X experience). This time, we dashed the bullet, but the conclussion is way too weak:
"If we still need the room, kick it out altogether"
This is not the first time, before we had another sign, canonical seriously considering replacing Rhythmbox (second most popular Linux music player, correctly working, good features, good amount of ubuntu users that like it) with Banshee (a feature lacking, mono-driven, unstable app only liked by Mono zealots) again under the extremely lame excuse of live CD size.
This situation is getting ridiculous.
Some points:
* The current default app distribution just works. Jaunty had tons of good reviews. In no place were there complains about X esoteric program missing.
* The current default app distribution already fits a CD. In other words, CD size is a problem only if you want to add new packages. What new packages? Is it really wise to replace raw functionality with some incredibly unknown package named "couchdb" that, I as an ubuntu user have never needed ?
[....]
59
votes
119
12
60
53
votes
114
4
61
Selected solution (#3):
Just use a DVD...
Written by
vexorian the 27 Jun 09 at 13:59.
Yes, that thing I said about DVD distros working in third world countries better than ubuntu is perfectly true. (I know this first hand). If the third world is not the reason for the insistence on using a CD disk for it then what is it?
Is it to keep the requirements low? That should be bull, really. We are talking about an OS for which most of the latest features (notifier thingie, compiz, ...) require a 3d graphics accelerator! Yet a DVD player is too high end!
Bandwidth? I do not think the live CD ISO would automatically jump to 2GB, it would likely start at ~700 for the first releases that use DVD.
Shipping costs? I don't think a DVD weights much more than a CD, I'd say the cost remains the same
ubuntu could just keep having a CD version ISO, that just ships the essential packages (Openoffice, The Gimp, firefox, gedit, brasero, totem, rhythmbox those are really the only apps you absolutely need, and I am sure that fits a live CD just as well)
Yes, that thing I said about DVD distros working in third world countries better than ubuntu is perfectly true. (I know this first hand). If the third world is not the reason for the insistence on using a CD disk for it then what is it?
Is it to keep the requirements low? That should be bull, really. We are talking about an OS for which most of the latest features (notifier thingie, compiz, ...) require a 3d graphics accelerator! Yet a DVD player is too high end!
Bandwidth? I do not think the live CD ISO would automatically jump to 2GB, it would likely start at ~700 for the first releases that use DVD.
Shipping costs? I don't think a DVD weights much more than a CD, I'd say the cost remains the same
ubuntu could just keep having a CD version ISO, that just ships the essential packages (Openoffice, The Gimp, firefox, gedit, brasero, totem, rhythmbox those are really the only apps you absolutely need, and I am sure that fits a live CD just as well)
-39
votes
26
19
65
Selected solution (#4):
Keep CD-Images for Xubuntu only
There are still old computers without DVD-drive (e.g. alot Pentium 3 computers). Ubuntu/Kubuntu might not run well on these but Xubuntu does.
If Ubuntu/Kubuntu uses bigger image sizes (>700MB), keep Xubuntu CD-Images for legacy support.
EDIT: Booting from an USB-Stick is also often problematic/unsupported on these old computers.
There are still old computers without DVD-drive (e.g. alot Pentium 3 computers). Ubuntu/Kubuntu might not run well on these but Xubuntu does.
If Ubuntu/Kubuntu uses bigger image sizes (>700MB), keep Xubuntu CD-Images for legacy support.
EDIT: Booting from an USB-Stick is also often problematic/unsupported on these old computers.
26
votes
40
14
14
Selected solution (#5):
Give greater preference to the network install
Written by
k33l0r the 4 Jul 09 at 08:24.
Make it easier to install Ubuntu using a 'network install' or 'FTP install' option. This way the latest packages could be downloaded directly from Ubuntu mirrors in the install phase and the user could select any extra packages that they may wish to install at the same time.
This would also mean that all packages would be up to date right after the installation.
An example of such an install process can be seen when installing Fedora.
Make it easier to install Ubuntu using a 'network install' or 'FTP install' option. This way the latest packages could be downloaded directly from Ubuntu mirrors in the install phase and the user could select any extra packages that they may wish to install at the same time.
This would also mean that all packages would be up to date right after the installation.
An example of such an install process can be seen when installing Fedora.
3
votes
16
5
13
Selected solution (#6):
Custom CD
I'd like to see the CD size mantained. They could try to keep squeezing things in ( that seems ideal for me! )
OR
We could have a link to a special website that would let you choose the apps you want to come with the live CD. Click Apply and then you would download an .img that suits your needs and saves bandwith to Canonical.
I'd like to see the CD size mantained. They could try to keep squeezing things in ( that seems ideal for me! )
OR
We could have a link to a special website that would let you choose the apps you want to come with the live CD. Click Apply and then you would download an .img that suits your needs and saves bandwith to Canonical.
12
votes
22
8
10
Selected solution (#7):
Target 1gb USB Flash drives
Written by
snadrus the 14 Jul 09 at 15:19.
These are down to $8 (Pricewatch) and more accessible to novices than CD burning anyway. They install faster and work on most PCs >512mb ram (an ubuntu requirement).
Rebuild the web page intending that people download an image and Windows/Linux/Mac USB creator.
Users can then use the USB drive for other uses.
For "older" PCs, soon 512mb will be "older" and everything will work except the CD drive (they usually break soon after the laptop battery). For these, Linux on USB is the best option.
These are down to $8 (Pricewatch) and more accessible to novices than CD burning anyway. They install faster and work on most PCs >512mb ram (an ubuntu requirement).
Rebuild the web page intending that people download an image and Windows/Linux/Mac USB creator.
Users can then use the USB drive for other uses.
For "older" PCs, soon 512mb will be "older" and everything will work except the CD drive (they usually break soon after the laptop battery). For these, Linux on USB is the best option.
0
votes
2
1
2
Selected solution (#8):
Install from repo on first click in the Application menu.
Written by
madhi19 the 17 Sep 09 at 12:10.
The first time you click on an Applications in the Applications menu it will install from the repo just like the Add/Remove menu. You could still keep a few "Must Have" like a web browser but everything else would be server dependent. That would make keeping a cleaner desktop even easier. Why should you even install the things that you never use?
The first time you click on an Applications in the Applications menu it will install from the repo just like the Add/Remove menu. You could still keep a few "Must Have" like a web browser but everything else would be server dependent. That would make keeping a cleaner desktop even easier. Why should you even install the things that you never use?
Solution #1:
Give Gimp, Inkscape and Scribus a unified look and feel
Written by
ubby the 26 Jun 09 at 10:14.
This will make Gimp, Inkscape and Scribus better to compete with the Adobe Creative Suite.
Because of a unified look and feel it will be easier to work with the programs.
To accomplish this the developers of the programs need to work together.
Maybe the developers can get help from universities who are working with graphic software.
This will make Gimp, Inkscape and Scribus better to compete with the Adobe Creative Suite.
Because of a unified look and feel it will be easier to work with the programs.
To accomplish this the developers of the programs need to work together.
Maybe the developers can get help from universities who are working with graphic software.
Solution #2:
All-in-one Modular Graphic Software
Written by
carvao the 26 Jun 09 at 15:45.
A software composed by different modules: a viewer, bitmap editor, vector editor, desktop publishing, web creator, and so on. The user chooses which modules wants to install or use by the time he/she opens a file or creates a new project.
A software composed by different modules: a viewer, bitmap editor, vector editor, desktop publishing, web creator, and so on. The user chooses which modules wants to install or use by the time he/she opens a file or creates a new project.
Solution #3:
Develop a "recommended" UI plan for gnome projects.
I'm not sure if one already exists, so bear with me.
One of the problems I believe is keeping programs from unifying is the lack of a recommended UI design. People start making projects and just throw buttons at it whenever they need a new feature. Over time the layout becomes caotic and a huge mess, so the project enters a "fix-it" stage where the UI is redone and made more functional.
The problem is that each project is doing this individually and only taking small hints from other projects. If there were a simple, elegant and extendable layout and scheme for gnome, or linux for that matter, it would help the projects in the "fix-it" stage to evolve into having similarly designed UIs.
What we need is that simple, elegant yet extensible layout design and a group of people (or projects) to start pushing it.
I'm not sure if one already exists, so bear with me.
One of the problems I believe is keeping programs from unifying is the lack of a recommended UI design. People start making projects and just throw buttons at it whenever they need a new feature. Over time the layout becomes caotic and a huge mess, so the project enters a "fix-it" stage where the UI is redone and made more functional.
The problem is that each project is doing this individually and only taking small hints from other projects. If there were a simple, elegant and extendable layout and scheme for gnome, or linux for that matter, it would help the projects in the "fix-it" stage to evolve into having similarly designed UIs.
What we need is that simple, elegant yet extensible layout design and a group of people (or projects) to start pushing it.
Solution #4:
Ensure that all creative applications work properly with Gnome-Globalmenu
Written by
ethana2 the 28 Jun 09 at 23:35.
When you have a large screen with a scribus or inkscape window maximized, a menu bar that's not on the top screen edge becomes inconvenient. While ubuntu isn't likely to default to a global menu for a variety of reasons, it's a crucial part of the professional workflow --ask any graphic designer who uses a Mac Pro with a Cinema display.
When you have a large screen with a scribus or inkscape window maximized, a menu bar that's not on the top screen edge becomes inconvenient. While ubuntu isn't likely to default to a global menu for a variety of reasons, it's a crucial part of the professional workflow --ask any graphic designer who uses a Mac Pro with a Cinema display.
Solution #5:
Introduce the core folks to each other
Written by
cyberix the 7 Jul 09 at 19:46.
Bring core developers of Gimp, Inkscape, Scribus, and maybe some OpenOffice folks physically into the same room to discus the matter.
Bring core developers of Gimp, Inkscape, Scribus, and maybe some OpenOffice folks physically into the same room to discus the matter.
Solution #6:
Introduce a unified tiling window manager
Written by
ethana2 the 9 Jul 09 at 07:56.
for dialog use by inkscape, gimp, and other applications. Few end users use tiling window managers on their desktops, but GIMP developers seem to expect them to. Speaking as an xmonad user, this is ridiculous.
for dialog use by inkscape, gimp, and other applications. Few end users use tiling window managers on their desktops, but GIMP developers seem to expect them to. Speaking as an xmonad user, this is ridiculous.
Solution #7:
Add the option to have similar keyboard shortcuts
I like each of the applications individually but it's a pain having to learn and remember a different set of keyboard shortcuts all of the time.
Adding an option to overwrite the keyboard shortcut for common tools (move, eyedropper, paint bucket etc) with a unified set would be very useful.
I like each of the applications individually but it's a pain having to learn and remember a different set of keyboard shortcuts all of the time.
Adding an option to overwrite the keyboard shortcut for common tools (move, eyedropper, paint bucket etc) with a unified set would be very useful.
Encourage torrents to download install CDs
Written by jmar71n the 20 Jun 09 at 11:01.
Related project: ubuntu.com .
Won't implement
Ubuntu should make the option for downloading CD's via torrents as a prefered option on the download page.
Currently there is only a small link in the bottom left corner.
Solution #1:
Use the new eye candy notification system instead
Written by
Apiman the 26 Jun 09 at 16:39.
Instead using a dialog window, use libnotify. It's much more beautiful.
Instead using a dialog window, use libnotify. It's much more beautiful.
Solution #3:
Use both notification and alert box.
Have the notification used when disk space is relatively low (10% for instance), but have an alert window (with action button) when disk space becomes critical (2 or 3% for instance)
Have the notification used when disk space is relatively low (10% for instance), but have an alert window (with action button) when disk space becomes critical (2 or 3% for instance)
Solution #4:
Change the color of the notification system for important things.
The notification system has to be different if for example a new song is playing or if the disk space is low.
So I purpose to change the color (maybe red), or make flash it.
It could be a great thing if the user should click the notification system, to show that he became aware of the warning.
The notification system has to be different if for example a new song is playing or if the disk space is low.
So I purpose to change the color (maybe red), or make flash it.
<a href="http://www.l2image.com/"><img src="http://www.l2image.com/images/x9ldsh8lfs1zgtq37vh.png" border="0" alt="L2Image" /></a>
It could be a great thing if the user should click the notification system, to show that he became aware of the warning.
Solution #5:
Tray icon
Written by
Lachu the 1 Jul 09 at 14:20.
Add tray icon called "show notification". In this mode user might read and interaction with notification.
Add tray icon called "show notification". In this mode user might read and interaction with notification.
Solution #6:
Notification logger
Written by
twocool the 1 Jul 09 at 21:31.
Create a daemon to log all notifications and a GUI application to see it.
Create a daemon to log all notifications and a GUI application to see it.
Solution #7:
Use Indicator not OSD
Written by
nachokb the 8 Jul 09 at 15:12.
Many of these proposal (including the screenshot) violate the NotifyOSD guidelines (no interaction, disposable, non critical notifications). For these kinds of stuff, I think the Indicator Applet is the appropriate medium. This was pointed at by cheesehead in the comments.
https://wiki.ubuntu.com/NotifyOSD#Interaction
Many of these proposal (including the screenshot) violate the NotifyOSD guidelines (no interaction, disposable, non critical notifications). For these kinds of stuff, I think the Indicator Applet is the appropriate medium. This was pointed at by cheesehead in the comments. https://wiki.ubuntu.com/NotifyOSD#Interaction
Solution #8:
Use a popup indicator
Written by
da brain the 8 Jul 09 at 22:45.
Use something like the update notifier that pops up from the top bar. It will flash to the user that it is running out of disk space.
Use something like the update notifier that pops up from the top bar. It will flash to the user that it is running out of disk space.
Make ubuntu more responsive
Written by marco.ferragina the 28 Feb 08 at 16:56.
Global category: System.
Won't implement
I've 2 distros installed on my laptop: Ubuntu Gutsy and Arch Linux! The feel with arch is that it's whippy. From the other side ubuntu is sluggish. Sure I know that my system is not so up to date (a 1.5 Ghz centrino 512 M Ram) but Arch work very well on it with the same services loaded as in ubuntu.
So I think Ubuntu can do something to improve performance.
2398
votes
2790
6
392
Solution #1:
Auto-generated solution of idea #85
Ubuntu Brainstorm was updated in January 2009. Since the
idea #85 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 #85 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!
261
votes
279
10
18
Solution #2:
Reduce the number of auto-started applications
Running "ps -eo pid,cmd,rss" provides interesting reading. A random entry is
4558 /usr/lib/gvfs/gvfs-gphoto2-volume-monitor 1576
although I am not currently doing anything with gphoto. One of the ideas of upstart was to start applications as a result of system events. While upstart is not the only answer here, I think that a few event daemons (the ones which are already running, like cron, upstart, inetd and hal are probably sufficient in many cases) could be used to start services when they are needed. A couple of generic gnome tray applets which could stand in for several existing ones (like indicator-applet is supposed to show notifications which were previously shown by a range of applets) and possibly start applications on demand would also help here. And all this would also help with reducing system start time as well as memory footprint of course.
I might add that there is probably some low-hanging fruit here, as in things that can be done writing little or no C code.
Running "ps -eo pid,cmd,rss" provides interesting reading. A random entry is
4558 /usr/lib/gvfs/gvfs-gphoto2-volume-monitor 1576
although I am not currently doing anything with gphoto. One of the ideas of upstart was to start applications as a result of system events. While upstart is not the only answer here, I think that a few event daemons (the ones which are already running, like cron, upstart, inetd and hal are probably sufficient in many cases) could be used to start services when they are needed. A couple of generic gnome tray applets which could stand in for several existing ones (like indicator-applet is supposed to show notifications which were previously shown by a range of applets) and possibly start applications on demand would also help here. And all this would also help with reducing system start time as well as memory footprint of course.
I might add that there is probably some low-hanging fruit here, as in things that can be done writing little or no C code.
232
votes
237
15
5
Solution #3:
Motivate or hire optimization experts to hunt for bottlenecks
Written by
leael the 17 Apr 09 at 05:26.
All of the following is related to "felt GUI performance", not 3D or other kinds of computational power.
Ubuntu (Gnome?) runs "OK" on my 600MHz/500MB/64MB, but I'd expect a 3GHz/2GB/256MB machine to be a lot faster, but it feels more like just 1.5 times faster.
Interestingly, the speed of CPU/RAM seems not to be in a linear relation (as it is with WinXP) to the felt GUI speed.
I remember, that X consumes extra time by passing everything through a socket, but that allone wouldn't explain the performance. I suspect, that there is much room to improve and do real optimization on many places throughout the system. Jaunty got much better, but it is still outperformed by WinXP.
So I suggest
to mobilize some real optimization experts and math geeks to look for possible improvements, and giving the developers hints, where to change what. If those experts are not to be found easily within the open source community, it might be a good idea to let hired specialists redesign problematic algorithms.
All of the following is related to "felt GUI performance", not 3D or other kinds of computational power.
Ubuntu (Gnome?) runs "OK" on my 600MHz/500MB/64MB, but I'd expect a 3GHz/2GB/256MB machine to be a lot faster, but it feels more like just 1.5 times faster.
Interestingly, the speed of CPU/RAM seems not to be in a linear relation (as it is with WinXP) to the felt GUI speed.
I remember, that X consumes extra time by passing everything through a socket, but that allone wouldn't explain the performance. I suspect, that there is much room to improve and do real optimization on many places throughout the system. Jaunty got much better, but it is still outperformed by WinXP.
So I suggest
to mobilize some real optimization experts and math geeks to look for possible improvements, and giving the developers hints, where to change what. If those experts are not to be found easily within the open source community, it might be a good idea to let hired specialists redesign problematic algorithms.
194
votes
207
12
13
Solution #4:
Make the GUI _seem_ to be faster, add responsiveness
Written by
leael the 17 Apr 09 at 05:42.
<rant>
On a Celeron at 2.6GHz/256MB GPU, I press ALT+Print, and have to wait 5 seconds (sic!) for just getting a tiny "Save As" dialog. That's just ridiculous.
</rant>
I have been using computers for a long time now, therefore I am quite quick on the keyboard and the mouse. With Ubuntu, my clicks can be sent to Nirvana, if I forget to wait for the system to be ready for new input. Such delays cost extra brain power and add up to frustration. It can even lead to damage through wrong input.
The overall performance of Ubuntu isn't really that bad, but certain things make it _look_ and _feel_ slow. Therefore another suggestion: Give the GUI a "virtual performance" kick by prioritizing certain key elements.
*) One example: Not knowing, if you really double-clicked that icon/gadget/element for seconds, is one of the worst failures imho. In this case, a visual feedback is needed immediately to let the user know, that the system accepted the command. I guess some part of the GUI must be run with higher guaranteed priority for this, therefore such feedbacks need to be kept at low weight.
*) Another example: Synaptics rethinks its software list for some seconds, after it had finished installing something. A click to the close button seems to have no effect. A kill -9 would do the low weight trick, just for picturing, what I mean; in this case the load routine could check for events while rebuilding its memory structures and leave the loop on a close command. Similar implementations should be considered for every loop, that can block an application for more than a (real) short moment.
<rant>
On a Celeron at 2.6GHz/256MB GPU, I press ALT+Print, and have to wait 5 seconds (sic!) for just getting a tiny "Save As" dialog. That's just ridiculous.
</rant>
I have been using computers for a long time now, therefore I am quite quick on the keyboard and the mouse. With Ubuntu, my clicks can be sent to Nirvana, if I forget to wait for the system to be ready for new input. Such delays cost extra brain power and add up to frustration. It can even lead to damage through wrong input.
The overall performance of Ubuntu isn't really that bad, but certain things make it _look_ and _feel_ slow. Therefore another suggestion: Give the GUI a "virtual performance" kick by prioritizing certain key elements.
*) One example: Not knowing, if you really double-clicked that icon/gadget/element for seconds, is one of the worst failures imho. In this case, a visual feedback is needed immediately to let the user know, that the system accepted the command. I guess some part of the GUI must be run with higher guaranteed priority for this, therefore such feedbacks need to be kept at low weight.
*) Another example: Synaptics rethinks its software list for some seconds, after it had finished installing something. A click to the close button seems to have no effect. A kill -9 would do the low weight trick, just for picturing, what I mean; in this case the load routine could check for events while rebuilding its memory structures and leave the loop on a close command. Similar implementations should be considered for every loop, that can block an application for more than a (real) short moment.
20
votes
35
5
15
Solution #5:
Auto-detect the appropriate apps to install
Written by
philip the 6 May 09 at 01:00.
There are trade-offs between performance on the one hand, and functionality and ease of use on the other. That's why we have several versions of Ubuntu, which install different sets of packages by default. This trade-off decision should be made (semi-)automatically. Given the hardware specs of the machine, the installer should suggest an appropriate desktop environment (such as Gnome or KDE versus XFCE versus Fluxbox), appropriate filer (Nautilus versus ROX), appropriate mail reader, web browser, terminal app, etc. The user should still be able to override those of course.
Likewise, the appropriate application settings can be made automagically for the trade-off. An example is previews in Nautilus.
There are trade-offs between performance on the one hand, and functionality and ease of use on the other. That's why we have several versions of Ubuntu, which install different sets of packages by default. This trade-off decision should be made (semi-)automatically. Given the hardware specs of the machine, the installer should suggest an appropriate desktop environment (such as Gnome or KDE versus XFCE versus Fluxbox), appropriate filer (Nautilus versus ROX), appropriate mail reader, web browser, terminal app, etc. The user should still be able to override those of course.
Likewise, the appropriate application settings can be made automagically for the trade-off. An example is previews in Nautilus.
-35
votes
5
4
40
Solution #6:
try lxde desktop
Written by
knb the 13 May 09 at 21:12.
Simply try the lxde desktop. It can be installed from synaptic.
Reduced resource consumption was one of lxde's design goals.
The can coexist in parallel on your computer. You can easily switch from gnome to lxde and back, at login time.
I have seen a very responsive knoppix system with lxde running on a tiny older eee pc.
Simply try the lxde desktop. It can be installed from synaptic.
Reduced resource consumption was one of lxde's design goals.
The can coexist in parallel on your computer. You can easily switch from gnome to lxde and back, at login time.
I have seen a very responsive knoppix system with lxde running on a tiny older eee pc.
8
votes
8
1
0
Solution #7:
Massive tweaks to nicity, priorities, scheduler, and limits
Real-time preemption:
So processes that live in the kernel (e.g. disk access) can't steal cycles away from UI code. (Unfortunately, this has *never* worked for me in getting the UI to be more responsive, but crashes brought down the system... maybe an expect could tweak it though?)
-15 nice-value for UI code:
So regular processes can't steal cycles away from UI code. (Unfortunately, nice has *never* worked for me in getting the UI to be more responsive. But maybe an expect could tweak it?)
Limits on disk access:
Automatically detect UI lag via a special program, and if so throttle disk access.
Disk scheduler:
From what I hear, CFQ seems "fair".
Process scheduler needs extensive review:
http://oreilly.com/catalog/linuxkernel/chapter/ch10.html
Cron:
Schedule cron tasks like updatedb to be run when the user seems to be away or sleeping, and stop when it detects the user returns to their computer. Namely, all non-essential services should be severely throttled at any sign of user interaction or CPU load. The current situation is *arbitrary* and can easily kick in a 100%cpu-and-entire-disk-bandwidth-hogging process of death in the middle of something important. Note that automated processes should not even kick in if the CPU is at >80% utilization even if the user is not at the computer (it could be a server). Also schedule tasks like updatedb to be run more rarely (monthly rather than daily).
Additionally compare notes with experts, other distros, and linux kernel devs.
Real-time preemption:
So processes that live in the kernel (e.g. disk access) can't steal cycles away from UI code. (Unfortunately, this has *never* worked for me in getting the UI to be more responsive, but crashes brought down the system... maybe an expect could tweak it though?)
-15 nice-value for UI code:
So regular processes can't steal cycles away from UI code. (Unfortunately, nice has *never* worked for me in getting the UI to be more responsive. But maybe an expect could tweak it?)
Limits on disk access:
Automatically detect UI lag via a special program, and if so throttle disk access.
Disk scheduler:
From what I hear, CFQ seems "fair".
Process scheduler needs extensive review:
http://oreilly.com/catalog/linuxkernel/chapter/ch10.html
Cron:
Schedule cron tasks like updatedb to be run when the user seems to be away or sleeping, and stop when it detects the user returns to their computer. Namely, all non-essential services should be severely throttled at any sign of user interaction or CPU load. The current situation is *arbitrary* and can easily kick in a 100%cpu-and-entire-disk-bandwidth-hogging process of death in the middle of something important. Note that automated processes should not even kick in if the CPU is at >80% utilization even if the user is not at the computer (it could be a server). Also schedule tasks like updatedb to be run more rarely (monthly rather than daily).
Additionally compare notes with experts, other distros, and linux kernel devs.
10
votes
10
1
0
Solution #8:
Email devs of other distros, asking what their tricks are for responsiveness
Send a very simple and quick email to devs of other distros, asking what their tricks are for responsiveness are.
Comments seem to indicate that other distributions do not have as much trouble as Ubuntu (and I'm inclined to agree).
Send a very simple and quick email to devs of other distros, asking what their tricks are for responsiveness are.
Comments seem to indicate that other distributions do not have as much trouble as Ubuntu (and I'm inclined to agree).
Annoying to send windows to another workspace
Written by edbian the 2 Jul 09 at 19:53.
Related project: Compiz .
New
Right now you have to either drag the window to another workspace (only with compiz if I'm not mistaken) and then move back. This takes some time and multiple clicks.
Alternatively you can right click and hit "send to workspace X". It simply disappears which is lame :) and it takes 2 clicks.
Solution #1:
Toss or Throw Windows using Compiz-Fusion
Written by
edbian the 2 Jul 09 at 19:53.
The general idea is that the windows have some basic physics. You can grab them (by the title bar or with alt + button1) and throw them around. If they hit the top or the bottom of the screen they should bounce off. If they hit the left / right they should slide onto that workspace.
1 click solution + eye-candy!
The general idea is that the windows have some basic physics. You can grab them (by the title bar or with alt + button1) and throw them around. If they hit the top or the bottom of the screen they should bounce off. If they hit the left / right they should slide onto that workspace.
1 click solution + eye-candy!
Solution #2:
Drag windows down to the workspace images
Written by
Anto the 5 Jul 09 at 09:15.
At the lower panel with images of the workspaces, what about being able to drag down a window and that way place them there? When you're dragging a window there, the window will get small so you can quickly place it into that workspace, and if you drag the window away from the lower panel, it will get enlarged again. After the window has been placed, you can select that workspace just as normal.
Less eye candy, but a whole lot more practical.
At the lower panel with images of the workspaces, what about being able to drag down a window and that way place them there? When you're dragging a window there, the window will get small so you can quickly place it into that workspace, and if you drag the window away from the lower panel, it will get enlarged again. After the window has been placed, you can select that workspace just as normal.
Less eye candy, but a whole lot more practical.
Solution #3:
Window caption should be optional
Written by
Lachu the 7 Jul 09 at 14:05.
When I hover mouse above Windows caption for some period of time, some additionaly options should be displayed, like: send Window to another workspace, force Window title, etc.
When I hover mouse above Windows caption for some period of time, some additionaly options should be displayed, like: send Window to another workspace, force Window title, etc.
Solution #4:
Make Expo plugin key bindings more apparent
Written by
euxneks the 10 Jul 09 at 16:24.
There is something that you can use to easily manage your windows on your desktop already, if you use compiz, called Expo. My key binding is e and it shows all desktops and allows me to drag windows between them. Make this a commong thing by either showing it in a tutorial, or allowing the user to right click on the workspace switcher applet to select expo.
There is something that you can use to easily manage your windows on your desktop already, if you use compiz, called Expo. My key binding is <super>e and it shows all desktops and allows me to drag windows between them. Make this a commong thing by either showing it in a tutorial, or allowing the user to right click on the workspace switcher applet to select expo.
Solution #5:
Don't change anything
Written by
gotham48 the 27 Jul 09 at 12:37.
Teach users to use Ctrl+Alt+Arrow
Teach users to use Ctrl+Alt+Arrow
Solution #6:
Add Workspace buttons to the title bars of windows
Written by
adamknox the 13 Oct 09 at 04:53.
I suggest adding buttons to the left of the minimize on the windows titlebar.
Because some people have named their workspace, this might not be the end all solution. I suggest these buttons be in place for up to four horizontal workspaces.
If there are only two workspaces it may make sense to use "send right" and "send left" images.
I believe that if there are more workspaces than that, or there are multiple rows that a single button should be added that, when clicked, displays a grid of the workspaces.
I believe that it should then that an option should be add in preferences to "follow" windows that have been moved to a new workspace, but that it should by default not follow. I think a nice little fade away effect could be implemented for those with normal or high graphics enabled.
I believe this is of utmost importance because workspace functionality is so unique to Ubuntu!
I suggest adding buttons to the left of the minimize on the windows titlebar.
Because some people have named their workspace, this might not be the end all solution. I suggest these buttons be in place for up to four horizontal workspaces.
<img src="http://i37.tinypic.com/xegcu8.jpg">
If there are only two workspaces it may make sense to use "send right" and "send left" images.
<img src="http://i36.tinypic.com/2dmb4wl.jpg">
I believe that if there are more workspaces than that, or there are multiple rows that a single button should be added that, when clicked, displays a grid of the workspaces.
I believe that it should then that an option should be add in preferences to "follow" windows that have been moved to a new workspace, but that it should by default not follow. I think a nice little fade away effect could be implemented for those with normal or high graphics enabled.
I believe this is of utmost importance because workspace functionality is so unique to Ubuntu!
Installer CD wastes time
No information about this blueprint
Information is updated every 5 minutes.
Please wait till the next update.
Written by korin43 the 9 Jul 09 at 09:32.
Related project: Live CD installer .
New
The installation CD isn't doing anything while it's waiting for user input for questions like locale/name. This time could be used to speed up the installation.
Solution #1:
Preload the CD while the installer asks questions
Written by
korin43 the 9 Jul 09 at 09:32.
Instead of copying packages from the CD to hard drive as soon as the installer starts, the live CD should start preloading as much of the CD as possible into memory while it's asking the user questions like their locale and name. This should give a noticeable speed boost on computers with enough memory (and shouldn't slow anything down on computers that don't - it would just stop preloading when it runs out of memory).
Instead of copying packages from the CD to hard drive as soon as the installer starts, the live CD should start preloading as much of the CD as possible into memory while it's asking the user questions like their locale and name. This should give a noticeable speed boost on computers with enough memory (and shouldn't slow anything down on computers that don't - it would just stop preloading when it runs out of memory).
Solution #2:
Search for the "net" while asking questions
Written by
Rodrigo the 12 Jul 09 at 15:38.
While you are answering questions the Setup program could be looking for networks, and maybe updating the packages list, so it does install the latest safety files, if the connection is fast, and the user wants.
Maybe asking:
would you like to download the latest safety updates, this could slow down the process of the installment in slower connections.
You wouldn't need to install something and then update it... it could be done in the same step
While you are answering questions the Setup program could be looking for networks, and maybe updating the packages list, so it does install the latest safety files, if the connection is fast, and the user wants.
Maybe asking:
would you like to download the latest safety updates, this could slow down the process of the installment in slower connections.
You wouldn't need to install something and then update it... it could be done in the same step
Solution #3:
Ask post-install questions during filecopy
Written by
snadrus the 13 Jul 09 at 16:27.
Slackware did this 15 years ago!
Questions:
, ,
Then while installing: (progress bar underneath)
, , , etc
If it 'feels' slow while you fight the CPU for copy time, then they could always wait until it's finished (like it is now). Obviously those options wouldn't commit until you "finished" the install. This could save 5 minutes.
Slackware did this 15 years ago!
Questions:
<language>, <keyboard>, <partition>
Then while installing: (progress bar underneath)
<user info>, <timezone>, <popcon>, etc
If it 'feels' slow while you fight the CPU for copy time, then they could always wait until it's finished (like it is now). Obviously those options wouldn't commit until you "finished" the install. This could save 5 minutes.
Solution #4:
Download locale files in background
Written by
tigr the 14 Jul 09 at 02:05.
As soon as user selects language (and if it differs from english) installer could start downloading required packages in background.
As soon as user selects language (and if it differs from english) installer could start downloading required packages in background.
Solution #5:
Simple Option: Read whole ISO image into memory and mount
Great idea. I'd like to propose a very simple implementation of it; that is possibly also the fastest/most efficient way to achieve this.
Given many install targets now have much more spare RAM during an install than the size of a CD; linear read the whole CD image in RAM, mount as an ISO, and install from there.
Linear copying the whole CD is the fastest way to get the install image into RAM and would normally only take between 2 and 7 minutes. So can ask all the simple install questions while that happens. The run the entire current install process from the RAM-mounted ISO.
This feature would only be enabled on targets with more than a CD-worth of free RAM (everything with 1GB or more?). On install targets without enough space RAM, just use the same from-CD install process as now.
Great idea. I'd like to propose a very simple implementation of it; that is possibly also the fastest/most efficient way to achieve this.
Given many install targets now have much more spare RAM during an install than the size of a CD; linear read the whole CD image in RAM, mount as an ISO, and install from there.
Linear copying the whole CD is the fastest way to get the install image into RAM and would normally only take between 2 and 7 minutes. So can ask all the simple install questions while that happens. The run the entire current install process from the RAM-mounted ISO.
This feature would only be enabled on targets with more than a CD-worth of free RAM (everything with 1GB or more?). On install targets without enough space RAM, just use the same from-CD install process as now.
Solution #6:
Don't do that at all
Written by
xfuser4 the 20 Jul 09 at 07:24.
There are several reasons why it is a bad idea to do anything in parallel before installation:
1. Loading from the disk during the user input phase of the installer could result in slow responsiveness of the user interface (at least on cheaper hardware, like my notebook, where disk accesses are resulting in a massive slow down of the system).
During the installation phase, Ubuntu should look very responsive - otherwise people will be disappointed.
2. Downloading locales or recent system updates during that phase is also a bad idea - at least, if you don't ask the user to do so. For several reasons:
- The user has an internet connection, that has a volume restriction (this is still usual in some places of the world). So you could destroy the quota of the user.
- To do any bigger access on the internet without notifying the user is generally a bad idea, because the user has not the impression to have the control over the system.
- Normally the user won't configure the WLAN settings, so this operation won't work at all.
3. The installation phase is the shortest phase of the system life time. Therefore it is a bad idea, to spend too much development time to it. The installation should work and should be user friendly - but no one really needs other fancy stuff there.
Ubuntu installs already very much faster than Windows Vista, Windows XP and MacOS X. The installation is clean and user friendly.
So it would be better, to solve other (more important) problems...
There are several reasons why it is a bad idea to do anything in parallel before installation:
1. Loading from the disk during the user input phase of the installer could result in slow responsiveness of the user interface (at least on cheaper hardware, like my notebook, where disk accesses are resulting in a massive slow down of the system).
During the installation phase, Ubuntu should look very responsive - otherwise people will be disappointed.
2. Downloading locales or recent system updates during that phase is also a bad idea - at least, if you don't ask the user to do so. For several reasons:
- The user has an internet connection, that has a volume restriction (this is still usual in some places of the world). So you could destroy the quota of the user.
- To do any bigger access on the internet without notifying the user is generally a bad idea, because the user has not the impression to have the control over the system.
- Normally the user won't configure the WLAN settings, so this operation won't work at all.
3. The installation phase is the shortest phase of the system life time. Therefore it is a bad idea, to spend too much development time to it. The installation should work and should be user friendly - but no one really needs other fancy stuff there.
Ubuntu installs already very much faster than Windows Vista, Windows XP and MacOS X. The installation is clean and user friendly.
So it would be better, to solve other (more important) problems...
Solution #7:
Dont Search for the "net" while asking questions
Written by
r3l1c the 21 Jul 09 at 18:36.
When I saw Search for the "net" while asking questions I immediately had my concerns. Connecting to the internet without the users permission is a Micro$oft problem that I have no desire to include within my favorite operating system.
I think that is a really bad idea
I do like the other ideas though
When I saw Search for the "net" while asking questions I immediately had my concerns. Connecting to the internet without the users permission is a Micro$oft problem that I have no desire to include within my favorite operating system.
I think that is a really bad idea
I do like the other ideas though
Solution #8:
Install to harddrive as if to a USB stick
Written by
kir360 the 24 Jul 09 at 17:53.
interestingly, the installation of ubuntu to a USB stick finishes very quickly. this might be because the CD image is directly being copied to the USB stick.
We can adopt the same method in here. the installation finishes and the system restarts. then a one-time menu pops up to configure the user details, etc. only the installation partition and grub path has to be given earlier.
interestingly, the installation of ubuntu to a USB stick finishes very quickly. this might be because the CD image is directly being copied to the USB stick.
We can adopt the same method in here. the installation finishes and the system restarts. then a one-time menu pops up to configure the user details, etc. only the installation partition and grub path has to be given earlier.
Solution #9:
A net install for highspeed connections
Add an option for an Internet inatall.
Add an option for an Internet inatall.
Solution #10:
Don't preload if the system is old
Written by
korin43 the 16 Aug 09 at 08:58.
Before starting the preloading thread, the live CD could check the clock speed and and amount of memory on the target computer. If it is below certain thresholds, the preload thread would not be started at all.
Where the lower bound should be is something that the Ubuntu developers should decide on, since they would know, but a simple example would be on a computer with 4 Gb of memory and a 3 Ghz Core 2 Duo, the Live CD would be loaded into memory (in the background). On a computer with 256 Mb of memory and a Pentium, it would not.
Before starting the preloading thread, the live CD could check the clock speed and and amount of memory on the target computer. If it is below certain thresholds, the preload thread would not be started at all.
Where the lower bound should be is something that the Ubuntu developers should decide on, since they would know, but a simple example would be on a computer with 4 Gb of memory and a 3 Ghz Core 2 Duo, the Live CD would be loaded into memory (in the background). On a computer with 256 Mb of memory and a Pentium, it would not.
Improve Flash Player support for Linux.
Written by Copitox the 9 May 09 at 17:18.
Related project: Flash plugin (macromedia) .
Implemented
Most of people are just "desktop users". They just want to listen to music, use an IM client, use twitter, facebook and youtube. So what? Well, youtube use flash, facebook use flash, and LOTS of sites use flash. It's clearly not the same wathcing a youtube video or playing a flash game in linux and in windows, the windows performance is WAY better. In Linux it barely runs, use a lot of CPU and sometimes even make the system crash. So, in order to make ubuntu really usabe for everyone, this should be solved.
527
votes
571
30
44
Selected solution (#1):
Lobby Macromedia/Adboe for better linux versions
Written by
Copitox the 9 May 09 at 17:18.
Canonical could use their connection with Dell to push Macromedia/Adobe.
Canonical could use their connection with Dell to push Macromedia/Adobe.
364
votes
411
42
47
Selected solution (#2):
Lobby Macromedia/Adboe to release the code
Written by
Copitox the 9 May 09 at 17:18.
Canonical could use their connection with Dell to push Macromedia/Adobe.
Canonical could use their connection with Dell to push Macromedia/Adobe.
325
votes
388
29
63
Selected solution (#3):
Work with Macromedia/Adobe
Written by
Copitox the 10 May 09 at 03:28.
Maybe Canonical could make a team to work with adobe on Flash linux versions.
Maybe Canonical could make a team to work with adobe on Flash linux versions.
353
votes
406
48
53
Selected solution (#4):
Improve Gnash
Written by
bitf the 10 May 09 at 03:38.
Gnash is Free Software, which means it can be bundled with Ubuntu. One step closer to getting Ubuntu to work out of the box.
Gnash is Free Software, which means it can be bundled with Ubuntu. One step closer to getting Ubuntu to work out of the box.
195
votes
254
42
59
Selected solution (#5):
Adobe Just Started, so just offer Help
Written by
Shady3D the 10 May 09 at 06:15.
just wait, because Adobe is just starting to develop for Linux, and they've spend a lot of time with windows to get this right, so its a matter of time.
so what should canonical do is just to offer Adobe for some help if they need, but working on Flash again will be waste of time and effort.
just wait, because Adobe is just starting to develop for Linux, and they've spend a lot of time with windows to get this right, so its a matter of time.
so what should canonical do is just to offer Adobe for some help if they need, but working on Flash again will be waste of time and effort.
61
votes
106
42
45
Selected solution (#8):
Improve SWFDec
Written by
anyedge the 15 May 09 at 07:16.
I list SWFDec(which actually works better for me) because Gnash has already been listed(and I voted for that as well). Adobe will only release their Flash code once the FOSS alternatives catch up to their software(think Sun and Java). Creating a team to work with Adobe WHILE they are keeping it closed source is a waste of time and resources that would be better served(and more coercive to Adobe) working on the FOSS alternatives.
I list SWFDec(which actually works better for me) because Gnash has already been listed(and I voted for that as well). Adobe will only release their Flash code once the FOSS alternatives catch up to their software(think Sun and Java). Creating a team to work with Adobe WHILE they are keeping it closed source is a waste of time and resources that would be better served(and more coercive to Adobe) working on the FOSS alternatives.
163
votes
167
9
4
Selected solution (#9):
Assist with HTML 5 (includes native video support, scalable graphics, etc)
HTML 5 includes native video support, scalable graphics, and other neato things. Basically the aim is to provide all the beautiful stuff we use on the web today without having to resort to some binary, unaccessible plugin.
--
When was the last time you saw a search engine tunnel through links in a Flash website?
--
HTML 5 is something that Canonical is able to assist with and support. Its a W3C, international standards kinda thing. It is where the web is headed.. or at least should head.Remember folks, the web got big due to accessible protocols..
--
We cannot let the future be run on closed platforms like Flash or Silverlight.
--
Think Javascript + scalable graphics + native video support all without plugins! Improve the open stuff, not beg the proprietary vendors for more. It's not the GNU/free way.
HTML 5 includes native video support, scalable graphics, and other neato things. Basically the aim is to provide all the beautiful stuff we use on the web today without having to resort to some binary, unaccessible plugin.
--
When was the last time you saw a search engine tunnel through links in a Flash website?
--
HTML 5 is something that Canonical is able to assist with and support. Its a W3C, international standards kinda thing. It is where the web is headed.. or at least should head.Remember folks, the web got big due to accessible protocols..
--
We cannot let the future be run on closed platforms like Flash or Silverlight.
--
Think Javascript + scalable graphics + native video support all without plugins! Improve the open stuff, not beg the proprietary vendors for more. It's not the GNU/free way.
56
votes
61
8
5
Selected solution (#10):
Work on ogv development
Mozilla and Wikimedia are working on the ogg video format to replace flash videos. This could be a viable alternative with a little bit of time and a lot of support from the community.
If someone can find the link on this, please post in the comments.
Mozilla and Wikimedia are working on the ogg video format to replace flash videos. This could be a viable alternative with a little bit of time and a lot of support from the community.
If someone can find the link on this, please post in the comments.
-22
votes
3
4
25
Selected solution (#11):
Do nothing
whats wrong with flash? the version from adobe works fine
whats wrong with flash? the version from adobe works fine
11
votes
12
3
1
Selected solution (#12):
Based on #4 Add GPU acceleration to Gnash
GPU acceleration will make flash media less of a resource hog.
GPU acceleration will make flash media less of a resource hog.
Professional-looking bootloader
Ubuntu
In :
Priority : Undefined
Definition : New (Needs guidance)
Implementation : Unknown
Assignee :
Mentorship is available if you want to fix this bug.
Written by Murrquan the 28 Feb 08 at 14:42.
Global category: Look and Feel.
Implemented
Ubuntu's bootloader is a stark black and white screen, filled with confusing options. It gives newbies a moment of indecision, as they try to figure out if they are supposed to choose something, and wonder why there are three or four Ubuntus listed. Then the timer finishes counting down (starting from 10), and the newb begins to feel like he's getting in over his head as his PC boots into Ubuntu.
Too much information up front, stark text-only display, painfully long countdown timer.
5944
votes
7511
21
1567
Selected solution (#1):
Create an attractive boot loader with grubgfx
Written by
Murrquan the 28 Feb 08 at 14:42.
Create an attractive boot loader with grubgfx that requires minimal user input and does not scare away new users. If that doesn't work then use Fedora's Plymouth
Create an attractive boot loader with grubgfx that requires minimal user input and does not scare away new users. If that doesn't work then use Fedora's Plymouth
524
votes
542
30
18
Selected solution (#2):
Use Plymouth and the latest GDM for a faster and smoother boot up.
Written by
vs8 the 14 Jan 09 at 15:59.
openSUSE and Fedora 10 feature more elaborate art than Ubuntu, I'm not talking about the colors they choose, because I like the Ubuntu colors, it's the art style.
One thing they got right is the boot up. Those distros boot very smooth and they look professional, Ubuntu looks rushed, plain and ugly.
I've seen Fedora boot, and the new plymouth thing rocks, Ubuntu should use it too. It's way better than the actual ugly usplash.
The other thing is the GDM, Ubuntu uses a very old GDM, which is slow. At least on my PC (AMD Phenon 9600 Quad Core, 4gb RAM). The new GDM is smoother, faster and it works fine.
In short, Ubuntu needs eye candy, from boot up to shut down it will attract more people, I guarantee it.
What is Plymouth?
http://www.phoronix.com/scan.php?page=article&item=fedora_plymouth&num=1
openSUSE and Fedora 10 feature more elaborate art than Ubuntu, I'm not talking about the colors they choose, because I like the Ubuntu colors, it's the art style.
One thing they got right is the boot up. Those distros boot very smooth and they look professional, Ubuntu looks rushed, plain and ugly.
I've seen Fedora boot, and the new plymouth thing rocks, Ubuntu should use it too. It's way better than the actual ugly usplash.
The other thing is the GDM, Ubuntu uses a very old GDM, which is slow. At least on my PC (AMD Phenon 9600 Quad Core, 4gb RAM). The new GDM is smoother, faster and it works fine.
In short, Ubuntu needs eye candy, from boot up to shut down it will attract more people, I guarantee it.
What is Plymouth?
http://www.phoronix.com/scan.php?page=article&item=fedora_plymouth&num=1
Screen:
https://fedoraproject.org/wiki/Image:Tours_Fedora10_012_Plymouth_Boot.png
openSUSE art:
http://www.opensuse.org/screenshots.php
New GDM login screen:
https://fedoraproject.org/w/uploads/9/94/Tours_Fedora10_018_Login_Screen.png
Thank you.
0
votes
0
0
0
Selected solution (#3):
Put a Motion Splashscreen on ubuntu
to better the user experience, ubuntu should have an animated Splash Screen. a (mini Video) playing, showing the Ubuntu Logo and then a smooth overpass from the splash screen to the loging screen.
to better the user experience, ubuntu should have an animated Splash Screen. a (mini Video) playing, showing the Ubuntu Logo and then a smooth overpass from the splash screen to the loging screen.
330
votes
349
19
19
Selected solution (#4):
Ubuntu Grub Backround
i was thinking that ubuntu team should design an default ubuntu grub backround for future releases..or atleast add them to the current ones. after all it would go nice after seeing the linux mint did that.
i was thinking that ubuntu team should design an default ubuntu grub backround for future releases..or atleast add them to the current ones. after all it would go nice after seeing the linux mint did that.
9
votes
9
0
0
Selected solution (#5):
Design a GUI to help with the process of changing the image
Written by
Rodrigo the 12 Jul 09 at 15:31.
Let's stay with the black screen or maybe a screen, but just one to begin with, then with the use of a GUI (the user can download it from the servers) be able to change it without needing to go to the console.
Everybody likes to customize his machine sooner or later.
Let's stay with the black screen or maybe a screen, but just one to begin with, then with the use of a GUI (the user can download it from the servers) be able to change it without needing to go to the console.
Everybody likes to customize his machine sooner or later.
516
votes
553
46
37
Selected solution (#6):
Use BURG with GRUB 2
Written by
tommis the 7 Jan 10 at 02:02.
BURG is a program based-on Grub 2,BURG uses new menu list,screen layout and shortcut keys that all easily customized.
<img src="http://img163.imageshack.us/img163/4537/screenshot300x221.png" alt="Image Hosted by ImageShack.us"/><br/>
Also see - https://help.ubuntu.com/community/Burg
-81
votes
50
15
131
Selected solution (#7):
Installation Option
In installation have an option of using either BURG or GRUB with GRUB the Default Option
In installation have an option of using either BURG or GRUB with GRUB the Default Option
48
votes
63
9
15
Selected solution (#8):
Just make grub2 better
Written by
lokster the 3 Feb 10 at 14:13.
The developers must make grub2 better. Why use another new and in early development bootloader, while there is already a good one?
Just make the good one even better.
The developers must make grub2 better. Why use another new and in early development bootloader, while there is already a good one?
Just make the good one even better.