R2 Desktop Proposal Rev. 2, Requesting Comments

Nice work :) and I like the idée’s .. this would really make Haiku an modern OS.

Morning.

DarkWyrm, first of all, thanks for taking the effort of writing the document. Now for some R2 ommisions / additions which will make Haiku stand out from the crowd:

1) Tracker.
- A GUI folder comparison tool (eg. which files are different between 2 directories and their subdirectories).
- Snapshot of all open Tracker folder windows and create "Snapshot Icon". Double clicking a snapshot icon restores all open Tracker folder windows on next session.
- The ability to Search for files on a non/BeFS image (even brute force will do).
- Each workspace has it's own Desktop folder, with different desktop icons for each workspace.
- Assign application to always open in specific workspace.

2) AppServer
- Group all common application windows to one master window with Tabs
- Dont minimise window to Deskbar, minimise to Desktop instead. Minimised applications actual appear like tab on side/bottom of screen. (can of worms - how are these minimised tabs related to Deskbar active programs list, and how do you show both?)
- Mouse over minimised tabs shows preview of currently minimised application.

3) Deskbar
- Introduce API which allows applications to overlay content on top of the program icon. This is quite useful for showing current application status.
- Introduce Deskbar API which allows applications to customise the context menu when the user right-clicks the deskbar programs icon in the list of active windows. See OSX Dock.
- Drag and drop new program icons to Deskbar/programs/category.
- Drag and drop FOLDER symlinks to Deskbar/bookmarks.
- Workspace specific deskbar, as well as master All-Workspace deskbar.

4) System application
- Encourage moving away from big-swiss-army-knife application philosophy to a smaller tool like application philosophy. A GUI tool similar to Cortex will then allow users to chain simple Tools/Apps and create Power Tools. Eg. image.jpg in => resizer => change colour depth => encode to png => image.png out

5) Apple have significantly cut down on the GUI animations. Leopard is almost dull compared to earlier versions of OSX. Yes, it was cool in early 2000's. It's annoying in 2008. OSX has almost come full circle from introducing transparency to removing it altogether. Linux/Windows have just started that game, and will come to their senses in 5 years or so. Haiku can benefit from completely avoiding this gimmick. I must admit that every time I boot to Haiku, I'm thrilled with its simplicity and purity, and enjoy an enviroment which is focused on speed and function, not lipstick and gloss. Other than adolescants, I believe that more and more people are sick of transparency on their desktop. YMMV.

Some idea i love, some i hate, some i'm indifferent (or i would need to use for a while to be sure of what i think of them).

Too bad you did not made it "frame maker" style (ie: 1.2.4 etc..) it would have made it easier to comment each point of interest. So i will take them by order with the title you gave to the section.
--------------------------------------------
1-Concerning 3D Acceleration:

That is a point you did not elaborate much, yet it is very complicated.
- I think it's needed because
Modern look bring users
On lower end PC, GPU can bring smoothness to the interface

- I think it's needed to still have a CPU-only way to do the compositing
We don't have much coder and being trap to a 3D API might remove flexibility of development.
CPU speed get better and geometry complexity of UI don't grow as fast as game, CPU might be enough soon. So if we can't code it fast enough, it's not worth it in the long term.

- I'm for vectorisation and even 3dfication (with 3d monitor), but i'm against "volumification"
Cubed workspace and tilted picture to give the impression of more screen estate, those are gimmick.
--------------------------------------------
2-Deskbar:

Here nobody will win. you say one way to organise, i will say another. That thing is personal and should remain flexible. One thing on witch we should agree is that a user A should easily find itself on user B computer. For that i think multi-user context would be the simpler solution, so we should forget the ultimate menu setup (one would be decided by default of course, but it should be as flexible to change it as current one is).
--------------------------------------------
3-System Information and Desktop Widgets:

Here i pretty much agree. What i could add is that i would like a better integration of this, something like info popper and the event sound pref panel. Also, it's ok to signal stuff to the user, but it's even better if the user can easily automate subsequent action in reaction to those info.

"Note that this applies only to the Deskbar and not the Desktop or other replicant trays.". This is stuff i don't like. I like Generalisation and i expect stuff to work and react consistently across a system.
--------------------------------------------
4-Easier Window Management:

Animated Hide / Zoom / Restore:
That will be a hard sell, of course it could be removable, but it's really bloat that add-up in the end. Not to mention i HATE that. I would be happy to have a clipUp 2.0 version without animation btw.

Show Desktop:
That is a windows feature i always felt was missing in beos. Of course i can go to an empty workspace or use "windows manipulator" (a bebits app/replicant) to acheive that. But a bring to front/back is more logical. In any case, i still find this a bad problem to think about because it's not really one fixable without causing other problem (so windows manipulator should do for now and close future). The real solution would be my old glassE sugestion to use a layer system for windowing just like in painting app. Not only it would bring the most powerfull and easy to use desktop around (everybody understand the layer concept) but app featuring layer would be even easier to code if the OS already have a basement for it.

Workspaces View:
In the context of accelerated desktop, this would bring very litte bloat and bring flexibility, i'm all for it.
Twitcher:
Nothing against that either. Only thing i would change about it is the ability to navigate it on 2 axiz instead of just a line.

Window Movement Constraints:
"window tabs cannot be dragged below the bottom edge or above top edge of the screen.". I don't like that. Moving windows freely is one of the best part of the concept as it make quick visual comparaison between 2 of them very easy and intuitive. I don't want to get old amiga feeling of windows somehow contained to the edges.

--------------------------------------------
Reduced File Management Effort:

Shift toward a tree-style file manager:
I like that, but i never liked the 2 panel view of those (it give me flashback of web site with frame). I think this must be to not stretch too much the tree while the selected part contain to many file, dunno. If that is the case i would pfefer a view with a slider bar when more than X files are present.

You don't seem to show attribute in your mockup, why? If it's becaue it would be a separate app to the tracker, i don't think having 2 "main" file manager is a good idea.

Search/Browse:
That is a very nice feature to have, not so easy to implement and still have it efficient. You bring a possibly good method and i'm not sure my own idea would be better. One thing i don't like is that your way go at the opposite of my generalisation philosophy (you split the find concept in 2). One other way to do that without spliting the concept would be to allow "level" of find (i prefer find to search, it's more positive :)). Level 0 would not exist, level 1 would be file, level 2 would be project etc... This is not obvious either, for exemple at what level would i put an archive like a zip?

One solution i once advanced on GlassE was to keep the file concept for backward compatibility and other OS interoperability but put the concept of Project at the root. The individual part of a project would then be hosted in a project and only linked from another project if common. As soon as a change would occur, only the diff would be saved in the linking project. If the original project is deleted, a subscription mechanism bring the item into one ot the project that need it. That Project concept would be the easier way to allow "contextual" find. The whole "context" would emmerge naturally from those project interrelation and linking of internal component. Of course, that would be more R3 meat than R2 :)
--------------------------------------------
Improving File Organization:
I agree with that, i find your app filer really good idea. Having it integrated into save panel is very good Vulcan logic :)

I think the best way to manage workspaces is not graphical at all but rather with a tree based manager. For example:
Workspace 1 internet (workspaces could be given a name)
-Firefox
--Haiku Project
--OsNews
-Vision
--irc.freenode.org
---#haiku
---#reactos
--irc.efnet.net
---#whatever
Workspaces 2 creative
-Wonderbrush
--a painting of a frog
Workspaces 3 office
-Gobe
--my business proposal
etc
This way you have a top down view of everything running. Shrunken graphics won't give you an idea of what's in windows with lots of text.

Very interesting reading, thanks :)

Apart from my comment about application menu, I'd like to couple of thought's I had.

Deskbar/System tray
I adore the BeOS way of having deskbar in the upper right-hand corner so that it can be under the window whilst tabbed title bar keeps its title visible! Keep it this way, make it be sort of hybrid deskbar, sidebar, taskbar, tray.
I appreciate your mockup is based on Linux system which I don't particularily like because none of the prominent desktop environments support vertical task lists. Even Windows can do this in relatively nice way.

Window management
Arrange workspace - neatly arrange all windows in current workspace, setting them side by side in a table (very much like Exposé but with actual windows not previews) based on their current location and shape, would work nicely with relatively few windows
Arrange visible - Same as above but addresses only topmost/non-overlapping windows; today in users of most systems have to arrange windows manually to have two or three different apps (e.g. html editor and browser with preview) side by side and then fine tune to make them to take advantage of maximum available/required space. Windows can do this (rather obscure feature though) but is limited only tiling and cascading. There is an option to do better :)

Twitcher
Haiku applications (almost) never use MDI approach so it would be sane to present all windows from all applications on the same level (without the need to switch from cmd-tab to up/down arrows). Users work with documents rather than applications so it's pointless to ask them to locate application first and then document.
What also might help is to indicate the screen location of currently selected twitcher item in order to utilise user's location memory.

From Filer to save-to queries
Take a step further from filer which automatically fills in the gaps. For example grabbing 'untitled' text file and dropping it to a query looking for 'type=text & name=my* & category=work' would automatically prompt for the '*' part only. Obviously this query would not accept images.
As per the mockup - the user could select appropriate query with attributes already filled in instead of suggesting a default name only. With attributes already filled in by query locating the file/sorting the dropbox later would be easy :)

some great ideas, but i think it would be in everyone's best interests to not perform any huge changes to opentracker.

Maybe give users the ability to switch between this new interface and the old tracker style desktop? (two separate code bases opentracker and "tracker2"?)

"Tracker2" could have all the fancy bubbly features and keep tracker compatibility.

Maybe I'm just too much of a BeOS fan-boy, I am still looking for the "tracker"-like Linux desktop :)

I just skimmed over the article, looks good so far. I didn't spot Internationalization (i18n) or locatization (L10n) or multi-user. As I understand it, these are planned for "future", but will they be R2 goals or R3 or later? It would be great if we can keep the scope of R2 short and focused and be able to go from R1 to R2 in a year or so.
-scottmc

I don't agree but I too found it to be annoying to click on an icon and accidentally triggering rename. One must click on the icon, not it's name. Will happen always when you are used to different behaviour.

But most of all I'd like iphone like pinch-resizeing of windows, that would be cool, Easy Resize.

Nicely done, but i have a few concerns about the Program Categorization.

Firstly no user without prior experience will know what "Vision" or "Beam" are. What are your solutions to a new user very easily navigating deskbar and starting a certain program?

IMHO I think the menus need to be more task-oriented instead of the less codename-oriented or even program name oriented.

Hi,

I have silently followed the Haiku adventure for a long time. Thank you (thousand times) to all the developers who made the old BeOS dream back on our computers.

I am not a haiku intensive user right now but there is one thing I really miss when using it : The ALT key modifier. Almost all the Linux window managers use ALT to grab and re size windows on the fly which greatly improves the windows manager experience.

Haiku: we love every bit of it.

@AlienSoldier, about project concept;
Yes, only I would call it a "bundle" for setting up relations between files , for grouping windows (action "bundle windows") and for aplication bundles like in mac os x. I suppose however you could manage files relations by copying or linking them to a folder, but this is 2008 and I want to see relationships and versioning in Tracker in the form of searchble attributes. A graph view in Tracker would really be "R2" as well :)

Good work, progress!

I hate to be a nag, and keep saying this
BUT ALL YOUR DOING IS REMAKING LINUX
Why not just leave it alone
Leave it like BeOS R5 was

A program menu that is broken down into categories
IE Games, Apps, Education, Office etc

Thats already been done, it's freaking called Linux,
And it sucks ass

My god, the software is free, and it is still getting its ass kicked by Windows. One huge reason is the stupid way of doing things

Most users dont have 50,000 programs. There is no reason to break them down like that. It makes Linux suck, it drives away users. It's ugly, it's unintuitive, it's just plain mucky

Instead of looking for a program in a list
You have to guess at what list it is under, and why, you wonder why it is under that list, when it could easily fit under another
Linux makes it worse, because so many programs names make no sense at all. But Haiku will suffer from it as well. And it will also make Haiku look just just another version of Linux. You do not need to compete with Linux, it makes no sense at all

I hate to sound unappreciative, but if Haiku has a categorized menu. I will not be using it. Not in any way, any how

Every time I install a new version or release of Linux. Thinking, this may be the one, this may let me finally get rid of Windows forever. The same thing happens. I see that god awful menu and format the drive. I dont even want to mess with it, format and rip it out. Not all because of the menu system, but because of that way of thinking

For years now I've been waiting for something, anything that could even come close to BeOS. I still have it running in my home, But I really just dont use it much, because it just cant do some of the things I do on a daily basis, like use flash drives

And for years now I have watched Haiku progress, thinking this just may be it. But now that it's getting closer to maturity, I'm realizing it's all gonna go to hell. In the end it's going to be just another release of Linux

The original be team didn't ask, what can we put in, they asked, what can we take out ? What sucks, what pisses off users ? What is a waste of processing power

Lets try a little of that,,, BOOKMARKS, not everyone needs them, not everyone uses them, they are for the internet, lets let the internet program of choice deal with them, WHY put them in the menu, why put them anywhere in the OS at all

There we are, problem solved,,, I still remember the first time I saw the new deskbar style in 5.0, I almost jumped out of my chair, holy crap, this thing remembers what I've done, this thing remembers documents I've opened, and programs I've ran, this is the reason I tried to break free from Windows

Dont they know thats why I'm using BeOS, because I dont like Windows watching my every move, because my kids dont need to see I just watched a porn movie ?,,, Now I have to go shut it off, remember to clear it,,, be annoyed by something that does not need to be there, believe it or not, I'm smart enough to remember what documents I've opened, and what programs I've run. The .008 seconds It just saved me from having to move my mouse a few inches, I could give a rats ass. If I want something like that, let me add it, let me download it, dont build it in, and make me learn how to shut it off

I hope you all know what I'm saying, and hopefully this will jog a memory in all of you. Haiku has the chance to be the nest BeOS, and you people have the chance to become millionaire's. And I have a chance to use a OS that doesn't do it all, that doesn't guess at everything. You are so close now, dont muck it up with thinking you know it all. Dont guess at what people want. If it is not needed, dont add it, make a widget they can download later

Dont try and build a better mouse trap, dont over do it
Make a OS that does nothing but work, and work fast

The next big shock came with the leaking of Dano
Once again, holy crap, the check in a check box, is animated
There a few more resources go, there is something I did not ask for or need. And may I remind everyone, Dano was not a release, they may have taken that out of the real release, had it ever been finished

There was really only two things I didnt like about BeOS
One was the deskbar remembering what you had opened
And two, was the animated check boxes in Dano

You have to realize, Windows has all that junk, you cant compete, you cant catch up, Learn what the Linux guys dont understand. Some people are trying to get away from it, not find a better way of doing it. If there was a better way. Windows would have done it

I thank god for all the hard work you people have done, and pray that you will make it, and be rewarded for your years of work
But the closer you seem to get. the more you seem to not want it to count for anything. Your over guessing it all, your changing what worked, your adding what is not needed

The squeaky wheel gets the grease, but the complainers and people who ask for stuff, thay ruin it for everyone else
Windows knows, ask the people for what they want, and 1000 people will ask for something that 10,000 people dont want

Anyone is free to make a widget, and anyone is free to download it and install it.

This is almost as bad as the people that want to get rid of the hierarchical folder system. it works dam it. We use it every day, in every file cabinet in the world. There is no better way. It's easy and it works

But people will say, there must be something better, and I'm just the guy to find it. I'm so smart that I will save everyone from sub folders, I'm a Linux guy

Yea, and 50 years later, they will still have 5% of the worlds users. We all know, the real power users, the real Linux gods, they use the shell, servers use the shell. Why ? because no one has over guessed the shell, no one has mucked it all up. It's raw power with no addons. And that people, was what BeOS was

If you want to fix it, if you want something better then Windows
Then take a look at what sucks in Windows, and fix those problems

Fix the fact that there are 50 ways to auto run something
And most users only know 1 or 2, to clean them out

Fix the most annoying problem in the world, that some programs dont uninstall correctly, or completely

Then, look at Linux, and fix what sucks about it

Fix the god awful categorized menu system

Fix the fact that there are 50 different ways to change 1 thing

Fix the fact that program names dont sound anything like what they do

Face it, one huge thing keeps Windows users from using Linux
Is the fact that Windows calls drives A, B, C, etc
While Linux needs a paragraph to work with a drive

Linux shows a file system that scares most users

Then look at Be, and how they managed to clean that up
To keep that part of the ugly file system from being seen
To make it look a little more like Windows

Oh yea, if there is anything in Be, that needed fixed
It was the right hand corner menu

Most people that will use Haiku, are people that read from top left to bottom right, it's the natural flow of things, it's only natural for the menu to drop down from the top left corner

But where are you guys on that one ???

Anyway, I'm sorry if anyone takes this wrong
And I wish you all the best

I sell thousands of computers a week, all around the world
If Haiku works like Be did, I'll be glad to put a copy on everyone of them. I still send 5 to 25 people a week to bebits and Haiku, just to show them there is a chance for something good to come.
I still have a BeOS machine running in our store Window, just for a conversation peace.

But if Haiku, comes with a Linux looking menu
Then I will finally bow down to Gates
And throw away everything I own with Be on it
Including the computers I have running it
You might as well call yourselves yellow tab 2

I totally understand where your coming from
But at the same time, I gotta compare it to the hierarchical folders mess.

There are really good reasons for wanting to fix it, but there really is no better way. The whole start menu thing at all these days is fading a little, with some people preferring to run portable apps. Way back in BEOS days, I really liked them better. I even used your tree application that way,,, always loved the idea of no install at all. And now people are loading portable apps on flash drives so they always have them handy, it reminds me of those days

I also know what it's like to want it your way
In BeOS 5 days, I used to drag the main hard drive over the the Be menu, then delete all the other links completely. When done, you dropped the menu down, and it listed the folders of the drive. So I had applications, media, documents, etc, and each one of them would expand out when you came over it

I loved it that way, cant remember how I got it working like that now, but it worked great. When I wanted to run an application, I just pulled the mouse over applications, then I could see each one as it expanded
Keep normal files out of root, and it looked and worked great,,, By the way, I would like to submit that idea for your thoughts should you really go ahead and make changes.

It gave you access to everything, the drop down menu with about etc, would list or show your own folders on the root drive, Drop down to documents to see your documents, drop to applications to see your programs, to media to see your music and pictures etc

Delete an application, and you dont need to worry about the link on your menu, because the menu shows the drive and files in real time
If you have never done it that way, you should really try it, after a week or so, I couldn't live with it any other way

But then again, thats what I'm talking about
No way, and everyone is ok, one way or the other, and 50% are pissed off

To bring something else up, if you dont mind
I remember talking to Scott hacker way back when
He really pushed the pro's of the Linux way of

This Folder, is different from
tHis fOlder

While I always argued that a real, normal, generic OS, should be free from that as well, A good OS should not care about caps, A better idea would be to make an OS capitalize every first letter, be it file or folder, and enforce it without asking, it makes everything so much easier and faster to read, and there is no real need to name 2 folders the same name,
Finally years later, he agreed with me.
That is a change I would not mind seeing in any OS

And lastly, while I'm on a roll
One other thing I would not mind, is not going overboard on file name length, back in the Win95 days, everyone went nuts with the idea that you should be able to name any file, any thing. At first it gave us freedom, later it gave us people using a small paragraph for a name, just try and find a file on bearshare, or kaza today, that doesn't run off into the next country and you would be lucky. A decent and normal limit of something like 32 characters would stop juvenile antics like that.
BeShare had some pretty nice people on it back in the day, but sometimes they went overboard also

Anyway, thanks for replying, I dont think I've spoken to you in about 8 years now. The kids came along, etc, and BeOS died out some, But I've tried to keep up and take a peek in on what you've been doing over the years

Now being so close to seeing something again, I felt I needed to get my 2 cents in, and I'm excited as hell that I may be able to get off my Windows machine again. 8 years, and hundreds of Linux CD's later, and I'm still using my Windows ME machine to write this. Just on a 64 bit CPU now

So to sum it all up, just to join in
Old menu, or live drive expanding
Say no, to Linux style folder/file names
Say yes, to forced, first letter capitalization
Put a respectable limit of file name length
Never ever, get rid of the BeOS CD inserted icon
Loved that thing, IE, instead of the MS autorun mess

You really gotta try expanding drive menu
I just wish I remembered how I did it. and it expands the way your talking about, IE applications expand to software name/company/manufacturer
Do it wrong and it looks weird

Exactly, good idea. I'm curious how it would work in practice, you'd see people files next to the vcard files for example in Tracker on a fat32 volume for example?

Well first of all let me say hello, this is my first post here. I've been reading through your document, and I can say we are at an exciting milestone here, the most of Haiku's future success depends on the decisions made here, so I'd like to share my thoughts, because I really care about where this project goes. I think that the greatest thing in BeOS was that it used all the resources in a small desktop computer, and it pushed out the last drop of every part it could. Nowadays we have multicore systems with very fast graphics cards, and other hardware parts. In a common BeOS sense, it's a crime not using them:) OS X is making the next step in Snow Leopard, they are now implementing OpenCL, which surely will be a great thing. But Apple's OS development has become a big slow giant mass, and that's where Haiku can (and should) gain advantage, all ideas can become reality without having to care much about compatibility with past technologies, so it's a really exciting moment here, maybe a next revolutionary OS will come out of the plans discussed here! The balance between using the GPU for accelerating the desktop, and using it's powers for advanced computations with technologies like OpenCL should be an interesting challenge. BeOS had a very good sense of resource management, this should stay in the future as well.

There is another thing despite using the latest hardware. All major interface advancements are defined by the path Haiku is targeted. I mean, it should be decided, where to go with the OS. BeOS was targeted as a mainstream media OS, all technologies were built for being one, get as much commercial media-related apps as possible. It's sad, but it couldn't go so far, it's history now.
Apple did a great thing here, they were strong in two major areas: they created hardware (so they wouldn't have to care about supporting hundreds of video cards and other hardwares), made deals with the biggest software development companies, and created free development tools for the developer community. Be did create a wonderful api, but as far as I know, they didn't came out with as sophisticated free development tools as Xcode is today. (but well, that was 10 years ago anyways.) So I think this is an area, where Haiku should really progress.

As of improving the user interface:
Be was using well proven professionals to create their GUI, it was well thoughtful until the last pixel. It's outdated now, no doubts. But we are talking here about extending some other people's work, so the only point I am here to make: be very very careful with every little change. All good things, the personality, the feel of the OS, the whole experience depends on this.
Maybe it would be wiser to come out with a whole different approach than the previous was, if the ideas represented in the new approach would be very-very revolutional, then people wouldn't be sad about losing the old one, but messing around with a legacy look and feel could easily end up offending peoples emotions. This community is very enthusiastic and emotional, so it's most important to keep them in love with the OS, because they are the ones who develop it :)

I will be following up all the improvements, and I wish you all a bright future in this project:) Keep up the good work, cheers

Peter

I see the proposal of animating hide/zoom/restore but is there anything about being able to resize windows in ways other than only the lower right hand corner?

Also, an obvious way to maximize a window to full screen.

Alan