Shell suggestion and gripes

[ml <ml arsis net>]

Like first I have to admit two things. I only tried livecd release so some gripes might not even be valid and gnome-shell is the best desktop experience so far. Can't wait for F15 (to my sadness fedora publishes other rpm repos only after release and I really don't plan on compiling everything I need, then again I finally found first reason to activate desktop counter). Strangely gnome-shell is much better and more to my likings than my tweaked desktop and offers exactly same features that I usually expect from my desktop, which are: being nice, clean, fast, without disturbances and most importantly different than normal broken design.

But not everything is perfect:
- Being avid gnome-do user, Super+type and run felt really good except the results being shown in icon view (personally I'd love possibility of selection between icon and list view for results). Results being shown in icon matrix made spotting and choosing with keyboard even harder. List view over icons any time for me, maybe it doesn't look as nice as icons, but it is sure as hell is easier to work with. Then again, maybe it's just me;)
- I found my self typing in search a lot more than in gnome-do which always remembered my favorites so i can simply press Super+Space,t,Enter and launches terminal (might be my failure since livecd probably can't store data)
- I use virtual desktops extensively and absolutely love being able to drop software icon on specific workspace (helluva better than searching them in icon matrix with keyboard arrows). Auto hide feature in overview is really unfriendly in this case often resulting with strange effect. Either being able to pin them to be always visible or GConf key for pinning them would be really nice. My most used computer is notebook with touchpad, just to clarify why auto hide is annoying
- Since virtual desktops are not static as before, but rather moving target I really missed Ctrl+Alt+End (to move to last and empty workspace) which could be simply followed Super and run
- Clicking on desktop without any effect feels downright strange. Are there any plans for that? I kinda want to write simple extension for mouse buttons on desktop to get standard quick launch menus (like in most Android launchers, where you can define what your specific keys function is) like Applications, Favorites, Places and so on and then user could decide which action he prefers. I'm just asking this so I wouldn't waste my time if 3.2 would fixate some desktop actions and make my extension pointless code

And as suggestion Contextual drop-ins

I'll try to explain the problem first as no desktop actually implements anything similar. In every desktop drag'n'drop is at best lame. Yes, you can drag some things, but it is as far from user friendly as possible. The less user knows about software and data, the smaller is the chance of him actually achieving any even the most basic thing.

Now, before anyone mentions d'n'd on dock (or d'n'd in shell overview). Please, don't. At best dock as it is, it simply resembles worst possible abomination when used in this scenario (not to mention that dragging data during overview is limited to shell and dragging data from software to overview would be insane). If you would want to put all launchers you want to execute or d'n'd on them you end up with absolute mess which is constantly present on screen. Completely different from this proposal and please, don't mistake. I love shell dock as long as it is containing only 5-8 icons i use the most. Every OSX computer and its dock abomination I see shortens my life by 10 minutes or so

Example cases: 
- If one is dragging folder named VIDEO_TS, there is a lot of chance he either wants to burn video dvd or use archiver to create iso file,... but i doubt he would like to print it, send it over mail or open it in browser, gimp or (god forbid) code editor/gnote
- If one is dragging picture or set of them, there is a big chance he would like to print them, open in gimp, post it on facebook, send over e-mail, print them... but i doubt there is any chance he would like to open in gedit or some even more incompatible software
- If one is dragging text selection from anjuta, there is a big chance he wants to post it on pastebin as c/python... code, store in tomboy/gnote... but it would be insane to think one would burn like to it on dvd or create video from it
- If one starts dragging contact, he surely won't want to drop it in dvd burner, he will either want to print contact, delete contact, send e-mail, contact over messaging....
- If one drags something.avi and something.srt, does he want to join these in one file? burn to cd/dvd? copy on media? but sure as hell,... he doesn't want to play solitaire with it

With simple d'n'd service providing only most basic messaging (and API) user could get smart proposals from desktop and thus desktop automatically became simpler for begginers. especially if drag-in toolbars are done in sane selectable and presentable way, like one for system and one for other tasks like messaging or social networking. there is no point in one drag-in for all usage cases at least in my eyes. If user learns social networking pops on right and system at bottom he will have much less trouble to recognize final effect as if actions are randomly stacked in one dock even if he is greeted with new icon in that part.

Messaging example:
1. on drag started application should simply invoke d'n'd service STARTDRAG(mime,data,otherpossiblemime) message and push it to registered drag-in docks
2. registered and user selected active drag-in docks show up
3.1. if dropped in invalid location service simply invokes ENDDRAG and docks hide away
3.2. if dropped in valid location there are 2 possibilities
3.2.1. drag location was satisfied with basic presented form of data and executes action while notifying service to promote ENDDRAG
3.2.2. drag location requested other form of data as specified in otherpossiblemime, and service again promotes ENDDRAG
4. drag-in docks hide as they were never there

The point of otherpossiblemime? simplest two examples, anything that is possible to present in pdf (or some other format) can be printable, thus printer is automatically valid destination. anything that is requested as trash can be deleted by dropping in global trash can (software receives request for trash, returns nothing and deletes dragged selection) and so on.

Every drop-in action would simply need to have 
- sanity check based on data presented in STARTDRAG if it is valid action for it and this decides if drop-in is shown or not.
- execution part
but basically it could be really simple scripts unless complexity is needed

Hopefully, idea doesn't sound great in my head only and I'm sorry if i spammed wrong list with my proposal;)

with regards
Matjaz