Slashdot Mirror
Improving GUIs In Open Source?
frekio asks: "I'm a 4th year undergrad at UCSD and am taking part of an ambitious cognitive science course that moves us, as groups, to redesign an interface with cognitive processes, usability and efficiency in mind. My idea was to contribute to Open Source with this project by assisting in a GUI design/redesign. We need to work with something that has an active user community which could provide feedback about changes we made or propose, and therefore must have a decently large userbase. What are some projects out there that provide software which people use day-to-day and could use more time and effort put into the User Interface?"
I agree!! I think dselect would be an excellent choice of things that could use a better design. It would be nice to have a graphical interface to dselect showing dependencies and stuff.
>~~~~~~~~~~~~~~~~
>~~~~~~~~~~~~~~~~
Pilchie
The install program for Debian!
*shudder*
I love the idea being Debian's design, but the installation of it gives me nightmares.
Add accessibility to GNOME. Right now, GNOME does not work well if you cannot see, for instance. All sorts of things need to be done to help out. Furthermore, there are loads of research into accessibility for GUIs to use as reference material. This wouldn't affect the average Dick or Jane, but would have a tremendous impact on the vision-impaired. Let's make Linux the preferred platform for the disabled!
--Be human.
Hehe, I have no argument that mice are required for playing first person shooters. The higher resolution the mouse the better!
I don't know that a mouse is essential for selecting things. Your specific example assumes using a Macintosh style file-manager application. I'm suggesting that we need to rethink both software UIs and hardware UIs. Perhaps file-managers only exist in the currently fashionable GUIs. Perhaps there are ways to select files that don't involve file globs or mouse-pointer-rectangles. I think the current UI model that we commonly called "desktop" is probably good for children and illiterates, but I suggest the desktop model actually impedes the activities of information workers. IT professionals could get their jobs done with or with a desktop operating system; let's have people examine what type of interfaces would be best for individual problem domains and put an end to the "one desktop gui" fits all non-sense.
I don't specifically attack keyboards, because they seem to be used, often in combination with other devices, by most UIs introduced to date. Mice on the other hand are very fashinable right now but aren't part of any UI except the desktop model. I'd be happy to be rid of the keyboard. Maybe I should just shut-up and start using my twiddler. ;-)
UIs could be a lot better if we all typed in Lojban instead of English. ;-)
Before problems with software user interfaces can be solved, we must solve the problems with the hardware interfaces. The mouse needs to be eliminated; it's great for slow people who don't know how to use a computer but it slows down those of us who know what information we want. However what the replacement would be is debatable. I like pen based interfaces (not for writing, for pointing) but that doesn't work well with the current fashion in PC 'stacking' (stacking the monitor on the desk, perpendicular to the desks surface).
Perhaps we need to ask the question, "do we really need to point in order to compute?" and if the answer is "yes" then how should we best point, and if the answer is "no" then what should be changed.
IMHO, using a wheel mouse to navigate and check out large numbers of pages / lists speeds up things immensely - at least for me. Having the Back button of the browser (and other apps like the Acrobat Reader) on the wheel (when it is pushed like a button) also helps. My image viewer understands the wheel mouse and is good for sighting large numbers of images.
I say that as someone who knows his accelerator keys very well - I still use them, but the mouse isn't for beginners only. However, apps must make sure that they can be used without a mouse.
The comparison of mice and keyboards isn't fair anyway because you shouldn't do the same things with both. There are jobs that require the shell exclusively, and vice versa (image editing). So, no simple answer here...
I was going to reply with the same suggestion.
The file dialog is the weakest part of GNOME and one of the reasons I use KDE.
Have you seen the KDE file dialogs? One of my favourite features is the bookmarks.
I also think there should be a home directory button as well as history in the text box.
However, it seems that the GNOME people want very much to be "not KDE." That was the point of GNOME in the first place, right?
It's too bad this question isn't on the main page. There is lots of good open source projects out there that would be that much better if they had a better UI.
The LyX project could do with some GUI design help, currently it is done by some rough guidelines of what is wanted, but some more top-down design and how things can be improved would be beneficial.
Disclaimer: I'm one of the developers on the project so I'm bound to be biased.
What about Gaim, Gnutella, or Gnapster? These projects have the advantage that they don't have many options/features. This means that the task would actually be possible to do for a class project. Bigger projects such as Mozilla and windows managers are complicated, have their own GUI libraries and if you aren't already a developer, good luck being able to come in and change the basic look-and-feel. Besides, I just want to see Mozilla 1.0, the last things that they need is somebody to change the interface for it.
Another suggestion is to work on the libraries. Sure you can do whatever you want to with a single application, but if you enhance the human-computer interface of, say, the gtk library, then you've made every gtk application better.
Keeping
A well designed version of dselect would be an ideal project. All the major hard work is done thanks to dpackage modules for python. You could use glade to easily build the gui, or ncurses to make a text interface.
While dselect works fairly nicely, I frequently screw things up in a big way because dselect expects too much of the user. For instance, instead of allowing the users to review changes to be made, it just makes them all at once. Sometimes something significant slips by when looking over long dependency lists.
A specific gripe I have is that sometimes it doesn't let me override the depenecie checkings system. At times I want to do this because I plan to install that particular component via a source build.
So, to summarize, redesigning dselect has everything needed for an ideal project. It has a horrid user interface to a complex process. It has the tools available so that anyone can attempt to replace it (ie dpkg-pythion), and it has a really large user base.
I'm a loser baby, so why don't you kill me.