Hi, Tangent fest ;-)
On 12/12/05, Linus Torvalds <[EMAIL PROTECTED]> wrote: > The reason I don't use Gnome: every single other window manager I know of > is very powerfully extensible, where you can switch actions to different > mouse buttons. Guess which one is not, because it might confuse the poor > users? Here's a hint: it's not the small and fast one. Just for the record, since I made this decision I can tell you that "might confuse people" was not the reason. More evidence for my point that "might confuse people" is the reason made up by others, not the reason given by the decision makers. First some context. The overall metacity plan was to first get all the defaults right as priority one, and then add more configurability and options consistent with keeping the defaults right. This was the driving "principle" if there was a principle at all. (The weekend I started on metacity the motivation was more "my [EMAIL PROTECTED] WM doesn't work, I'm just going to write one that works how I like") On the specific feature of arbitrary button bindings, the full discussion is archived in bugzilla. But my memory of this feature is: - I put in a lot of special cases to get the default behavior exactly right; the event handlers for mouse buttons do not look like "run the action associated with this button," they are more complicated - I spent a few days trying to code a patch that made button actions configurable while preserving all the detailed behaviors I had coded, and I just kind of gave up because the patch was too hard/complicated/big and I wasn't willing to break the default behavior in order to simplify the code. - I did put in configuration of the most common stuff people wanted to change, like double click action and alt+click modifier key, and this made most people happy (based on reduction in bugzilla/email traffic) My patch is still in bugzilla, if anyone wants to start from it and find the simple and elegant way to code it. The patch as I left it is buggy though and had a couple "hard to fix" problems. Plus it's against a pretty old version of metacity I guess. BTW, though I confess that I like to reject window manager patches, I also spent a ton of time getting EWMH usable and supporting it in GNOME. The only purpose of EWMH is to make the window manager replaceable. You may be noticing that I like the idea of "choice of two well-focused designs" better than "single choice of one nobody-hates-it design." Anyway. The primary issue with preferences in metacity was never confusing users - that would only be an issue with displaying prefs in the dialog, i.e. unlimited prefs would be OK, as long as they were hidden. The more important issue I always had in my mind was the quality of the defaults, and ability to spend time polishing the defaults. The tradeoff came from amount of personal time I had, code complexity, and interdependencies among prefs. But, I pretty often flamed people complaining about lack of prefs in bugzilla, so I can't really whine about being misunderstood :-P > Same with the file dialog. Apparently it's too "confusing" to let users > just type the filename. So gnome forces you to do the icon selection > thing, never mind that it's a million times slower. I don't think "too confusing" was the reason here either, though I can't speak authoritatively since I didn't design this. There was also a bad rap here since in the original design spec (and current file selector) you can in fact just type the filename. The text entry box appears as soon as you press a key. You can also press Ctrl+L to get a text box with autocomplete. But version 1.0 didn't have this since the coders ran out of time. I'd also point out that OS X makes the same basic decision as GNOME to avoid the "foo/bar" path notation in the default UI, so while it (agreed) is not ideal for users who are primarily shell users, I don't think it's a particularly radical or unprecedented choice in the big picture. Havoc _______________________________________________ Usability mailing list [email protected] http://mail.gnome.org/mailman/listinfo/usability
