Linked by David Adams on Tue 9th Dec 2008 16:46 UTC, submitted by weildish
OSNews, Generic OSes Sometimes it seems that every new OS that comes out these days is ultimately and altogether quite similar. "It's all been done before," we sigh as new system after system is released with only eye-candy the apparent difference for most users. This new OS, named "g-speak" by its creators, will give one a run for his money. Using special gloves, a user of the system gestures his way about the OS on several wall-sized displays that interact with one another. The makers of g-speak call it "the first major step in [a] computer interface since 1984." Perhaps they are right. See the neat video here. You can even go so far as to dust off your Tom Cruise Minority Report action figure to better savor the future with.
Thread beginning with comment 339724
To view parent comment, click here.
To read all comments associated with this story, please click here.
RE[2]: New UI not OS
by rajj on Tue 9th Dec 2008 20:50 UTC in reply to "RE: New UI not OS"
rajj
Member since:
2005-07-06

What exactly is an object oriented file system supposed to be? Where you have abstracted containers with common methods to act upon them-- sort of like... files?

As far as their contents being application or document centric, that's really up to the applications and libraries used to access it. The file system proper doesn't --and shouldn't-- really come into play.

Reply Parent Score: 3

RE[3]: New UI not OS
by Morgan on Tue 9th Dec 2008 21:16 in reply to "RE[2]: New UI not OS"
Morgan Member since:
2005-06-29

More information about OO file system concepts:

http://www.usenix.org/publications/library/proceedings/coots97/full...

Reply Parent Score: 2

RE[4]: New UI not OS
by rajj on Tue 9th Dec 2008 23:39 in reply to "RE[3]: New UI not OS"
rajj Member since:
2005-07-06

Well. That was the point of my comment; you made it sound like it was obvious what you were talking about as if there were some formal definition of an object oriented file system or that it was fundamentally different from what we already have. Now that you've shown me an actual implementation, it makes more sense.

Having a way to add user-space plugins to the VFS layer to override vnode syscalls per file based on type gives the file system a polymorphic feature which is indeed _one_ property of being OO but hardly one that OOP has a monopoly on. That said, this is still a neat idea, but I'm not convinced it would be as revolutionary as you suggest.

It also seems like every pre-existing programming concept gets lumped under the OOP umbrella. If it exhibits even one of those features, then the OO hand waving begins. I've even seen literature for products brag that the system is OO because they call UI elements --yes you guessed it-- OBJECTS!

Reply Parent Score: 3

RE[3]: New UI not OS
by Morgan on Tue 9th Dec 2008 21:26 in reply to "RE[2]: New UI not OS"
Morgan Member since:
2005-06-29

As far as their contents being application or document centric, that's really up to the applications and libraries used to access it. The file system proper doesn't --and shouldn't-- really come into play.


This statement shows you don't understand the difference at all. The current way of doing things is that the application decides what to do with the file. Conversely, I think the user should decide what to do with the file, and the OS should open the correct application based on the user's decision. This is facilitated through the file system; in my example above, when a user selects a file he is presented with the possible tasks for that file based on the file system's understanding of how the file can be manipulated.

An example: Do you want to print a rich text document? Instead of opening TextEdit or Word, then opening the file, then choosing Print from the menu, you should be able to just click once on the file, then select Print from the actions available. You've just saved yourself and your computer a great deal of time and processing cycles.

Reply Parent Score: 2

RE[4]: New UI not OS
by Bill Shooter of Bul on Wed 10th Dec 2008 17:34 in reply to "RE[3]: New UI not OS"
Bill Shooter of Bul Member since:
2006-07-14

An example: Do you want to print a rich text document? Instead of opening TextEdit or Word, then opening the file, then choosing Print from the menu, you should be able to just click once on the file, then select Print from the actions available. You've just saved yourself and your computer a great deal of time and processing cycles.


I'm not sure if you are saying that is a hallmark of some futuristic file system, or what not, but that's more than possible today. Take windows XP. Right click on a rtf document and then chose print from the actions. It doesn't involve the any file system metadata currently. Not sure how it would improve the situation.

Reply Parent Score: 1

RE[4]: New UI not OS
by DrillSgt on Wed 10th Dec 2008 19:20 in reply to "RE[3]: New UI not OS"
DrillSgt Member since:
2005-12-02

"An example: Do you want to print a rich text document? Instead of opening TextEdit or Word, then opening the file, then choosing Print from the menu, you should be able to just click once on the file, then select Print from the actions available. You've just saved yourself and your computer a great deal of time and processing cycles."

You have been able to do this for years now. It is the right-click versus the left-click.

Reply Parent Score: 2

RE[4]: New UI not OS
by looncraz on Fri 12th Dec 2008 06:40 in reply to "RE[3]: New UI not OS"
looncraz Member since:
2005-07-24

I don't see, at all, how the file system would even matter here. Seriously, it is the UI that changes, nothing else matters, to the end user.

For Windows, Explorer, MacOS X - Finder, BeOS/Haiku - Tracker. By whatever name, it is a program that gets changed. Indeed, much of what is being described is already implemented in the only currently feasible ways, considering the OS can't handle all data natively - allow applications to customize Explorer / Finder / Tracker options / or install add-ons for more than just capability reasons ( think patents & anti-trust ).

The interface is the only change, and merely adds more steps to the work flow ( which ain't always bad, but is normally ).

For instance, if I have an "object" (file) on the Desktop or in a folder, and I click on it to open it up or use it for some purpose, a list suddenly fanning out would be bad. There would need to be a pre-requisite for that action - such as double-clicking it, or right clicking. The action list would merely be a menu displayed in another manner.

The "OS" would need to require or rely on the applications, capable of doing something with a given file, to register themselves in a database so that the action can be performed. This is true regardless of the file system. Indeed, the file system would only be useful for performance improvements - and that depends on the implementation.

The actions that could be performed would depend entirely upon the plugins, extensions, or applications installed in the system. Some tasks would be common.

In the end, all you have, is what we already have - with a different look & feel ( which might not be a bad thing ).

For instance, I have some homework - an essay, I need some text input and tools - which means a program to edit the file. There are only a few things I can do with text, but there may be numerous programs on the machine capable of doing it. I double click my essay, and I get a list of actions - Print, Edit [ plus many common options ]... the more options, the worse the interface. ( all already possible in Windows/Ubuntu, with a right click context menu )

For instance, given an image there are only a finite set of unique generic operations: View, Print, Edit. And they aren't so unique, but not all files can be edited.

The end result will simply be to the following effect:

CommonFileActions{ // known as "object" to the UI
delete,
rename,
email,
encrypt,
compress,
etc...
};

SpecialFileActions{
edit = true,
print = true,
view = true
};

That is about it. Files have limited actions which can be applied to them. The only thing we can hope for is a standardized set of open file formats so that operating systems are expected to handle those natively, without relying on third party applications.

And, this all still applies even if the Desktop model itself is destroyed. Something will represent a movie, and there is no valid reason to change view to play for a movie - I mean NONE.

Otherwise, allowing special actions such as "Spell Check" "Blur Image" "Sharpen Image" "Send Virus" just clutters the interface while providing too many options for the user.

Indeed, the only great change in the UI will come when computers are thought controlled, or have a specific special application or placement ( such as on a wall ). Nothing will change too far beyond what we have now on the fundamentals - regardless of how that is relayed to the user. There just aren't enough valid reasons for such a change to occur.

--The loon

ps: I guess I should also mention I am only speaking of the end-user perspective :-)

The true OOP system is merely the implementation in a singular system - a single API, of what already exists. But that plays well into the open file format needs, where the system can treat all images the same way, regardless of format. BeOS can do this, but not all apps on BeOS take advantage of it, nor does Tracker have any reason of doing anything particularly special because of the capability.

Edited 2008-12-12 06:55 UTC

Reply Parent Score: 2