Don't most distros let you click on a .deb or .rpm and have it popup a 
dialog asking for the root password, telling you it's going to install it?
Yes.  Most probably do.  In writing this I was thinking I should probably have a play with the current state of Nautilus/RPM integration.  Though if I remember rightly from the last time I tried it, Nautilus just popped up redhat-config-packages and tried to use that to install the RPM.  Not quite what I was looking for.  The RPM and DEB package management systems are designed around unattended installation.  They don't plug into GUI's that well and they certainly don't provide ways of displaying a "readme".  If I remember rightly it is specifically against policy for an RPM to issue any output _at all_.  Overall I just thought the MacOS way "looked nice".  Also, I like the Applications folder as start menu kinda thing.  One huge EXE is the entire app etc... I think the ROX desktop uses a similar idea.

Part of it is that applications aren't "documents", and people often 
think of files == documents.  Thus people don't often seem to realise 
that applications actually exist in the filesystem. (Speaking to people 
in the Windows/Linux world anyway).

This is true and begs the question do "files" for user documents make sense.  Personally I'm starting to lean in the direction of databases for user data.  Users shouldn't have to organize their data on their computer or even understand that one "document" is stored in a different file to another.   They should also not have to explicitly "save" what they are working on.  I use the IntelliJ Idea Java IDE and while it does have an explicit save option you don't need to use it.  If you perform any function in the IDE that would need the data on disk to be in sync (compiling etc) it saves for you.  It also saves the data when you switch windows or whatever.  I've never lost any code for any reason, even when idea crashes (which it does rarely).  If I need to revert to a previous copy or I don't want to retain some change I've made I can just use the history (kinda like local CVS) or undo.

The whole separation of volatile (RAM) and non-volatile storage (disks now, was tapes) needs to go the way of the Dodo. It was born of an era when memory was extremely limited and insanely expensive.

The whole idea of applications starting and quiting is also a bit old school.   Something else I noticed with MacOS is that an app will quite happily continue to run if you close it's main window (this has been the case on Macs for eons).  It can take some getting used to the application menu at the top of the screen all the time.  I'm so used to windows having their own menus and toolbars and quiting when you click the little X on the last open window.  One thing this causes you to do is forget to quit apps.  To quit an app you have to press Apple-Q (or choose it from the menu) but I don't see why you should ever have to quit an app.  I'm so used to doing it tho I find myself "cleaning up" my running apps every now and again on the Mac.  Given enough ram and swap space it just shouldn't be required.  Switching between apps should have same UI as 'starting' one (and with the MacOS dock it is).  The computer can tell if an app is running or not and transparently to the user start it when they want to do something that requires it.

I have 1GB of Memory in my work PC.  This is entirely affordable these days.  Rarely even with huge Java apps like IDEA running do I exceed 60% usage.  The rest is filled with disk cache and buffers.  I don't see RAM getting more expensive and the exponential growth in memory capacity has not been met with an exponential growth in memory usage by applications.  Database style 'document' management may change this but I don't think it will to an excessive extent.

Regards 




-- 











Oliver Jones » Director » oliver@deeperdesign.com » +64 (21) 41 2238 

Deeper Design Limited » +64 (7) 377 3328 » www.deeperdesign.com