We can't make Linux "better" and "ready for the desktop"-- which does *not* mean we have to do everything via a GUI, dagnabit; people can certainly use the command-line comfortably *if they know how*-- unless we identify where people are falling over it and how to remove the obstacles to their understanding and ease-of-use. Difficulties using error output effectively looks like an obstacle to ease-of-use to me.
Unfortunately, Holly, I don't think linux will be ready for the desktop for quite awhile (yes, that does make me sad).
Blah, blah, blah-- no offence, but we've all heard this opinion and the reasons for it many times, and frankly, I'm tired of it. Tired enough to make an effort to see if it's actually *true*, rather than just assuming it is. Tired enough to try to cross the gap between what the vast majority of computer users have been force-trained to accept/expect, and a real alternative computing experience.
Heck, an auto is a replacement for a horse, but they are by no means the same. New, alternative technology does take a while to begin to be accepted, but once it proves useful, mass acceptance often comes rapidly.
This is the part where I find the whole argument falls over somewhat:
Nobody wants to take linux in the direction of windows (thankfully), and since most of the linux developers are power-users they have no reason to want or include this kind of brain-dead junk in their software.
I understand this, but it assumes that there are no alternative routes, and that makes me just nuts.
Everyone assumes that "the masses" must have a GUI, because the command-line is "too scary" in some way.
But if the command-line was intuitively understandable, *would these users still be scared of it*? Would they continue to avoid Linux, just because text is not as "pretty" as icons, even when the text was just as "easy to use" as the icons, and the system as a whole had many other benefits?
Maybe, maybe not, but the only way to find out is to *actually try it*-- and that starts with asking real people who experience problems just what those problems were and attempting to determine the problem's root cause, so we can find out just what needs to be adjusted to help future users cross the gap.
I'm flatly sick of assuming that the only choices are to turn Linux into a Windows clone as a "bridge", or leave the user stranded on the shore strewn with "brain-dead junk". Can the gap be forded? Can we teach the user to swim? Can we provide a raft-- or wings?
-------feel free to stop reading here; useless debate follows---------
Because of the wide use of windows any replacement OS (be it linux, bsd, macosx, or whatever) would have to function in a similar way before it would be accepted. The following would be a base set of requirements for such a replacement:
1. Boot totally into a gui - no startup output. Those messages are great for someone trying to diagnose an issue, but are just confusing to some and unnecessary to most, which is why windows boots to gui and totally hides this kind of information.
Yes, so I have a bootsplash. So does Windows. So do many distributions install this by default.
2. Totally configurable via gui - no low-level file editing. As power users this is something that we want/need, but the windows user expects to pull up a dialog for the program and click checkboxes to turn things on and off. I can just imagine the dialogs necessary to configure something like postfix or sendmail ;-)
Not necessarily true. Yes, younger users who only have experience with Win95 and above expect this, but you know, all of us aren't young, and those of us who used Windows for more than 10 years well remember DOS and Windows 3.11, and editing *.ini files-- which in many ways was more effective than trying to manage the bloody Registry. Heck, those who use alternative shells under Windows are familiar with configuring "low-level" files without a GUI. People create themes, people set up web pages, people do all kinds of things under Windows that require them to deal with low-level (or intermediate-level) configuration and commands. So I'm not completely convinced that this "one click to rule them all" business is a truly accurate portrait of the "average" user's experience/knowledge level.
Secondly, the ability to actually *configure* said server might well overcome the fact that it's "low-level file editing", assuming the low-level file was understandable to edit (which brings me back to where I started). If in fact you can run a mail server on Windows (I have no idea), how is it configured? Via the GUI you're imagining necessary for Linux? If so, it has been done once, so can be done again. If not, and it's one of those things that Windows calls "hands off", as it does with so much, then my point is made.
Thirdly, if you're running a server of some sort, then you ought to have a bit of familiarity with the necessities of that server insofar as the "low level file" should hopefully not be as obtuse as it might be to Joe/Jane Average, who is likely not running a server anyway. In which case, we're again talking aesthetics, which I postulate should not be a primary argument as everyone makes it. It's an important thing, but not the *most* important thing.
3. Less service-oriented and more interactive. Sure we run ftp servers, web servers, mail servers, etc. And we expect them to go off and do those things without bothering us. But at this point the windows user expects visual feedback on everything - a mail icon indicating there's new mail in outlook, blinking network light showing network activity, other tray icons with menus allowing you to get to the background 'services' right away.
Got most of those. Kbiff mail checker, KNemo network monitor. Since I'm in SuSE atm, I can use YAST to deal with the background services that aren't in my tray (like the New Hardware monitor, which is in my tray). Mandrake would have the MCC. RedHat would have some useless piece of trash that I couldn't understand (that was my experience with RH). We can all use LinuxConf or Webmin. Gnome users have gnome-system-tools. And there's no reason someone couldn't provide a bunch of desktop icons that ran various start-stop-restart scripts, except that no one has bothered to do so.
4. Self-updating. M$ has been pretty poor in this respect but they are actively working on it and getting better. My windows box downloads updates automatically, installs them with a nice progress bar (and not a lot of detail), and either a) handles whatever is necessary to get the new updates used or b) asks me to reboot for the changes to take effect. The whole process is totally brain-dead, and that's what the average windows user is going to expect.
Many Windows users don't like this, btw, and often don't use the auto update. Windows updates are often unreliable. This is why Microsoft is going to force users to accept updates whether they will or no, in the near future. However, many Linux systems are capable of this as well; certainl SuSE is (SuSE Watcher), and many systems without distro-specific tools have cron jobs set to perform various updates in the middle of the night.
I don't agree that the "average user" necessarily expects or in fact desires a totally brain dead process. That's usually what got them in a mess in the first place, and many users, though "average", are more than clever enough to recognize that.
What I'm wondering is, if the needed processes were not "easy" per se, but were nonetheless simple to 1) recognize, 2) understand and 3) perform (none of which necessarily requires a GUI base), would they still be so objectionable that such users would avoid Linux nonetheless.
Everyone seems to assume that they would be, but I am not convinced that that is the case. But we would have to first (find out how to) make such processes simple to 1) recognize, 2) understand and 3) perform, and then see how user perception responded.
Holly -- gentoo-user@gentoo.org mailing list
