This started in another thread relating to having a better GUI. Some of us argued that other issues might have higher priority.

I am collecting comments for a possible poll to see what people think are the most urgent issues with Linux. Please submit possible items to go on the list--in a short format similar to that below. (Feel free to expand also)

Also, any other thoughts.....

To start the list:
(not in priority order)

1. better and more consistent printer interfaces
2. better video drivers
3. less confusing and convoluted xorg.conf
4. Professional features in GIMP--eg color profiles
5. A Browser that acts like IE---to deal with brain-dead webmasters
6. more robust wireless setup tools, drivers, etc.
7. universal boot loader which auto-detects all installed OSes and does not required a config file.
8. ONE mainstream retailer who puts a Linux computer on display

For the distributions to follow LSB

and

Open source drivers
Open source drivers
Open source drivers

oh..wait I forgot something else

Open source drivers

I'd really like to see a standardized USB device support. On some distributions all I have to do is plug in my thumb drive and off I go. On others I have to recompile the kernel. And then on others I have add udev rules. Its just painful. I'd really like to see it "just work" on all distributions. I think thumb drives are becoming "standard equipment" for a lot of people. Just like I shouldn't have to mount my hard drive every time I turn on my computer, I don't really feel like I should have to mount my thumb drive manually every time I use it too.

I would like to see a desktop thats designed around the same principals as the command line, rather then a crappy desktop abstraction. Its a computer, not a desk or foot rest.

I would say wireless is the main issue at the moment. Its been how long since wireless authentication and encryption been around and were still using convoluted methods to get things working ndiswrapper madwifi intel ( Intel psychosis should be a better term) its just crazy. Take intel wireless (not the ones in the kernel ) you have to compile IEE80211 stuff out of the kernel compile the driver move firmware over compile a damon and then compile wpa_supplicant to get anything to work and after all that maybe 40% of the time it all works. to get it to work you have to load the driver load the daemon and then start wpa_supplicant....

xorg.conf I guess it could be better but I just wish they did not go to a full modularized xorg package modularizing the drivers was a good idea but not the whole X org package.. If you ever compile it from scratch you feel the pain of xorg 7x



Printer interfaces.... yeah never had an issue with this cups is pretty good at this stuff

of course ive never had a usb printer before.

Not an issue with Linux but more with gimp


I do not think that this is even possible to accomplish.

I say this because of the limited amount of computing power available at this stage in the booting process.

I think the biggest obstacle here would be for the person
trying to implement this is the vast amount of different
hardware that Linux can run on, and the variations of
possible concurrent OSes they'd have to install, test and
maintain against their product. I certainly would feel that
cost AND effort are prohibitive.


Cheers,
Tink

Because at some point Pixellany has to turn these into a poll, could I ask that you all do your bit to help by doing this:

• one liner for the poll about your request

Now, you may be wondering why I want this. Firstly, it makes life a damn sight easier for Pixellany. He doesn't have to spend time trying to figure out what you want and then try to condense it into a one line poll option. It also makes it easier for us to see what you plan for your idea.

apologies for singling out the "open source drivers" one, but it illustrates perfectly what I'm on about

I would like to see hardware vendors provide drivers for linux as loadable modules with a basic front end to use the device. Or some kind of tool to convert windows drivers to linux drivers.I doubt this would ever happen.

Hey Vendors! Did you know that windows is not the only operating system on this planet?

My one liner for your poll:

Better co-operation between vendors and the OSS community.

A side note, since i started using my first distro i've noticed a huge improvement, i just want it to carry on :P

All;

Thanks for all of the good input. I am going to setup the poll this weekend, so this is the last call for inputs.

Remember, we're just collecting topics here and not starting a debate. I'll probably keep 1 or 2 that have already collected flak, but I'm guessing they will not score high.....

I want to reiterate my comments in the thread that spawned this thread. I understand that the purpose of this thread is to "brainstorm", but we need to think ahead...

So far everything in this thread is too nebulous to provide any valuable input to anyone. Sending an e-mail to Alan Coopersmith at Sun that says, "We need a less confusing and less convoluted xorg.conf" is going straight to the trash. We need to provide concrete requirements to the developers. FLOSS projects are no less engineering development projects than development projects undertaken by corporations using paid employees. Engineering projects must have requirements to have any chance of being successful. The more well-defined those requirements, the greater the chance of success.

I would suggest that we provide the most detailed requirement we can in this thread and let pixellany et. al. distill them to a one-liner for the poll (I will help you pixellany, if you want). Spend some time thinking about your perceived problem and at a minimum answer the following questions in your post:

• What is the current condition?
• What is wrong with the current condition?
• Why is it beneficial to make the change?
• Who benefits from the proposed change?
• What is the required, final condition?

For a very quick examples that pixellany can use if he wants...

Currently many FLOSS projects use GNU autotools to generate configuration scripts. Some projects edit the "template" configuration script to include a summary of the successful configuration. Many projects, however, do not summarize the configuration. These summaries are helpful to ensure that the final installed product includes all of the functionality desired or required and excludes undesired functionally. Summarizing the configuration is more efficient and less error-prone than reviewing configuration logs or screen output. The "template" configuration script produced by the GNU autotools should provide a summary including, at a minimum, the following information:

• Environment variables that are set; e.g., CFLAGS=-m64 -mtune="k8" -O2, PKG_CONFIG_PATH=/usr/lib/pkgconfig
• Options passed to the configure script that control compilation/installation. Especially when they override default conditions; e.g., --prefix=/usr, --libdir=/usr/lib64
• Dependency options that were passed to the configure script and their status; e.g., --enable-gtk-doc=yes --without-x=yes

Arow;
Good points.

At this stage, I don't worry too much about the details of how to present a problem to the potential fixer. If something like X-windows configuration pops up to the top, then the right people can get the right problem statement written.

The key in my mind is illustrated by this vastly over-simplified assertion: "Please don't spend more time on fancy GUIs and eye-candy. I want better print drivers and more robust video driver install and configuration."

This said, you and XavierP and other are right that we need to think ahead to how we actually solve these issue.

I heard my name I think that we need to have the statements nailed down here. To, again, pick on the "open source drivers" solution, at first glance that seems really very reasonable and something we can all get behind. But if that were your task, where do you take it? Do you contact the kernel devs, the xorg devs, the printing devs?? or do you send out a mass mail to every hardware manufacturer you can think of??

As far as I am concerned (and if this isn't your aim, please correct me), we need to start a worthwhile dialogue. Let's say that a problem is that the Gnome interface is too simplistic. And yes, I am channelling Linus here! The steps are simple: work out a statement of the problem, get onto the Gnome mailing list and post up your statement asking for debate and discussion, profit! Even if the outcome is that nothing changes, we will have forced someone to explain why in plain language and we will have maybe effected a change. We will also have something to report back.

Think we got this one 99% already. I generated a config file for booting the majority of the 145 systems before I install them.

A few distro installers did have a go to compile a configuration to boot all the systems I have in the PC.

Technically nearly every Linux installer is capable of detecting the existence of an OS but it is a lot of work to decide on what kind of Linux. In the case of a MS system, which has a common MBR until Vista, it is not possible to tell a Windows from a Dos by checking just the boot sector.

I don't know if it is possible to fit this in 512 Bytes, but on the MBR there could just be something that checks every primary partition to see if it starts with a certain sequence of bits. If it does, it takes this as meaning the bootloader is on this partition and loads it. That partition could be on any primary partition and be as large as it likes as long as it has the required sequence in the required location. Then the boot loader can be as large as it likes and on any primary partition is likes.