fabianr wrote:I would suggest to actually push only one aspect of activities at the moment:
Act when the (hardware) enviroment changes:

User plugs in a mouse, keybord, external monitor on a laptop, ask the user if she wants to use a different enviroment for workstation setup.

User plugs in a second monitor, ask the user if she wants to use a different enviroment for dual monitor setup

User plugs in a tv, asker her if she wants to use PMC

...
maybe give the option to switch activity on a specific wlan?

cool idea. I like the google questions: do you want to use xxx save selection as standard.

andreas_k wrote:

fabianr wrote:I would suggest to actually push only one aspect of activities at the moment:
Act when the (hardware) enviroment changes:

User plugs in a mouse, keybord, external monitor on a laptop, ask the user if she wants to use a different enviroment for workstation setup.

User plugs in a second monitor, ask the user if she wants to use a different enviroment for dual monitor setup

User plugs in a tv, asker her if she wants to use PMC

...
maybe give the option to switch activity on a specific wlan?

cool idea. I like the google questions: do you want to use xxx save selection as standard.

While I like the idea as well, this is exactly the kind of confusion with converged form factors I mentioned in my earlier post. What you're saying makes sense from an end-user's perspective and this type of workflow is probably desirable. The problem here is that there is a distinction between Activities and Workspace type (like Desktop, Netbook, PMC and Active). When your hardware environment changes, you could want to only change your Workspace type, e.g. when you plug in a TV, ask if you want to use the PMC Workspace type (this would implicitly be in your current Activity). You could also want to change to an entirely different Activity, e.g. when you plug in a TV, ask if you want to use a different Activity which may be better configured for the PMC Workspace type.

I think people could use either of these workflows, so both of them should be possible. But it is important to clarify when you're only changing the way things are presented to you (only the Workspace type) and when you are changing the way things are configured (only the Activity) and when you are doing both.

I would suggest to switch to a new activity. If you want to just switch your workspace type one could clone the activity and change the workspace type.

Correct me if I'm wrong, but at the moment "Desktop" is the only workspace that is considered stable anyways? PMC is beta, and if one switches to PMC one would use it a "single fullscreen app" pretty much all the time anyways?
Maybe I have missed it, but I haven't read anything about Active or Netbook in a long time.

If you clone the activity and change the workspace type when you just want to switch your workspace type, that will only work the first time you clone it. Any changes to the one activity after cloning will not be available in the other activity. So this has a different effect than just switching your workspace type.

I don't think it matters how many stable workspace types are available right now. The point here is to avoid confusion when additional workspace types become available in the future. While some people may want to always switch activities as well as workspace type at the same time, this is not the case for everyone. So you shouldn't enforce this. It also doesn't work with the basic usage of converged form factors.

One of the basic use cases for convergence (as i understand it) is as follows (note that this is unrelated to the Activities functionality):
- I'm using Plasma on a laptop/tablet hybrid with a Desktop workspace type in Activity Home
- I'm working on a document in Calligra
- I'm tired of sitting at my desk and disconnect the tablet part so i can continue working on the document on my couch (possibly because I'm already at the point where I only have to review some stuff and don't need the keyboard anymore)

At this point, I should only change my workspace type and not move to another Activity, because then the document will not be shown. If I clone the Activity, the document might be shown again (not sure about this), but this would only work the first time you clone the Activity and might not keep the parts of the document that aren't saved yet. Even if you clone the current Activity every time, you end up with way too many Activities so this doesn't seem like a good solution to me.

Thanks for bringing up the topic of the effect of the environment (Available monitors, wlans, bluetooth devices, ...) on the way your computer works. It is an important feature for the future but I would not include this in the discussion of Activities / configuration / Virtuals Desktops. The concept is orthogonal: Plugging in a laptop may want you to switch to media center activity or disable screen turn-off in all activities - so it may effect only one or all activities, one one or all virtual desktops. Imho this environmental effect needs to be considered when designing the future configuration (e.g. system or desktop settings), so I would suggest to bring it back in, when our discussion reaches that point.

For the names, I see a slight tendency for workspaces or environment for activities, but keeping Virtual Desktops as terms? Do you agree? Any better idea to intuitively name the concepts (which we all seem to pretty much agree on)?

I had a look at KDE Personas and tried to imagine how they are interacting now with virtual desktop and activities, what type of issues they face and what could be done to improve their experience. I focused on Philip, Santiago and Matt, as I think Susan is not organised enough to use either virtual desktops or activities, and Berna might use VDs but not activities. Hopefully these concepts can be made easy enough to be used by both of them. I am not at all a UX expert, so I may be misusing these personas only to prove my point Sorry if this is the case…

Philip uses extensively virtual desktops and see no point in using activities as he is used to his workflow and he thinks activities are getting in his way. He uses only VDs, but he chose to associate each of them with one of his activities (top right VD for administration, the bottom left for videos, etc...). He even gave them a name to reflect that. In advance KWin settings, he also used rules to make VLC always launch on his bottom left desktop.
Philip uses some advance features of virtual desktops which relate to MM1, not to MM3. I think a lot of confusion comes from this overlap: activities were added as an additional layer to virtual desktops, while some of its features were already included in VDs. Instead, I think each tool should be specialised to fulfil only one Mental Model. Activities should be the reference tool for MM1, and virtual desktop should be refocused to only fit only MM3. This means removing the features which are related to MM1 from virtual desktops. I suggest that virtual desktops would be considered only as a way to group windows, nothing more. Giving desktops a defined spatial positioning gives the idea that they are persistent. To reinforce this idea of groups of windows, desktops should be only a one dimension list (no more top-left-bottom-right), where you can reorder the desktops if you want, easily add new desktops, etc... To go even beyond that, I think adopting a behaviour similar to Gnome-shell, where a desktop is deleted as soon as the last window on it is closed, would be a good idea. Of course, this would break Philip’s workflow, so activities should be improved to enable him to do all what he was already doing, even more comfortably.

Santiago is also a user of virtual desktops, he likes to use them to organise windows, but unlike Philip, he does not use advances features. He actually started using virtual desktops when the desktop grid effect was introduced, as it helps him have a spatial representation of desktops. He does not use the pager to switch desktops. He can easily send a window to another desktop. He tried activities, but could not really understand how they could benefit him. He did notice that right clicking on the title bar of a window enabled to send it to another activity, but he did not understand at a first glance what was changing when he switched between two activities, and just did not bother to try to understand or look for documentation on this feature.
The desktop grid effect helped Santiago to have a spatial representation of desktops. Activities come as a third dimension, and I think that it is easier for our brain to work in 2D on a 2D screen (psychologists in this thread might confirm or infirm this guess of mine). Activities are displayed in a vertical list in the activity manager. Maybe activities could always be presented in a vertical list, while desktops would be presented in a horizontal list. Screens on your desk are usually positioned side-by-side, this would be coherent with desktops (screen-extensions) as horizontal. Having an effect to organise windows in desktops and activities would be useful. I made a suggestion of how this effect could look like in this thread.

Matt took the time to understand what activities were about, he played around with it, looked on the internet how they could be used and finally he set up two activities: one for his geology courses, and one for all the other things. These two activities are pretty separate: he mainly browse the web, watch movies and play games when he doesn’t work, and he needs an office suite for working, a software for managing his stones pictures, etc… However, in both cases, he likes to listen to music and see his last emails. In his student activity, he often has to manage many windows, and use a lot virtual desktops to sort them. In his “home” activity, most of the time he is focussing on only one task at a time. A few months after starting using activities, as he was looking for something else in System Settings, he realizes that he can define an activity-specific power management profile. He finds this very useful but would have liked that he had discovered that before.
First, Matt needs a lot of virtual desktops in his “student” activity, but only once in a while he needs a second one in his “home” activity. Having a different number of desktops in each activity would make sense for him. Generally, having a hierarchy between activities and desktops might help to better understand both MM1 and MM3.
Also, he needs different applications, but his favourites in Kickoff and his launcher in IconTask are the same whatever the activity. This should be changed.
Then, even if Matt uses activities, he didn’t know that it could also affect the power management profile. Currently, there is no central place where you can configure all the settings which are related to activities. I think every one of them should be included in the Activity Manager, this would help a lot in understanding MM2. In System Settings, default settings are defined. In the activity manager, for each activity, you can choose to either use the default, or override the settings with something activity-specific. If a setting is overridden, a warning is displayed in the corresponding System Settings section, “this default setting is overridden in the current activity”, and links to the corresponding section in the activity manager. All activity-specific would be centralised in a common place, which would improve discoverability. Also, people like Philip who don't want to use activities would not be bothered by activity-specific settings in the main System Settings.

bjoernbalazs wrote:Thanks for bringing up the topic of the effect of the environment (Available monitors, wlans, bluetooth devices, ...) on the way your computer works. It is an important feature for the future but I would not include this in the discussion of Activities / configuration / Virtuals Desktops. The concept is orthogonal

I understand that the concept is orthogonal, but that was precisely why I wanted to bring it up at this point. I think we need to keep in mind that this feature will be coming in the future and we should rethink the Activities accordingly. It would be like telling the people in the past that were creating the Virtual Desktops feature that the Activities feature would be coming in the future in some form or another. Maybe they wouldn't have named it Virtual Desktop and we wouldn't have to avoid that confusion now. Of course, Activities probably weren't even on planned back then so it's not strange that they couldn't take that into account. But at this point we already know about the future features relating to converged form factors and environment changes, so we should take them into account, e.g. by making sure we're not using terms that are better suited for describing form factor changes or other hardware changes.

bjoernbalazs wrote:For the names, I see a slight tendency for workspaces or environment for activities, but keeping Virtual Desktops as terms? Do you agree? Any better idea to intuitively name the concepts (which we all seem to pretty much agree on)?

I also see this tendency, but it seems to me like workspace is used more for what we describe with Activities, namely the entire desktop, including virtual ones, and its configuration. I think environment is used more when describing the hardware and peripherals, sometimes also including the workspace. I think Virtual Desktop can be kept, to avoid confusion but also since it's not actually an incorrect term so there's no reason to "correct" it.

AGuiFr wrote: Instead, I think each tool should be specialised to fulfil only one Mental Model. Activities should be the reference tool for MM1, and virtual desktop should be refocused to only fit only MM3. This means removing the features which are related to MM1 from virtual desktops. I suggest that virtual desktops would be considered only as a way to group windows, nothing more. Giving desktops a defined spatial positioning gives the idea that they are persistent. To reinforce this idea of groups of windows, desktops should be only a one dimension list (no more top-left-bottom-right), where you can reorder the desktops if you want, easily add new desktops, etc...

I don't think you should make a hard split between the functionalities provided by Virtual Desktops and Activities. I think the overlap in functionalities allows some flexibility on the user's part. If they need only a little bit more functionality than the hard-split Virtual Desktop would provide, they need to start using Activities which might require more effort. This could discourage the user from using that extra functionality, because it's not worth using Activities. On the other hand, if you have that functionality available in Virtual Desktops, at the point where you outgrow the needs of the Virtual Desktops it should already be worth using Activities, despite the effort required, since Activities should be able to perform the same functionalities much better and with more flexibility.

AGuiFr wrote:The desktop grid effect helped Santiago to have a spatial representation of desktops. Activities come as a third dimension, and I think that it is easier for our brain to work in 2D on a 2D screen (psychologists in this thread might confirm or infirm this guess of mine).

I agree that the Activities come as an additional dimension, but I don't think this needs to be changed. I think that that's the whole point of Activities, having another layer of configurability for your desktop. But it may indeed also be the reason why it's so difficult to explain clearly. Perhaps it would be useful to have a concept demo video showing a mockup of how a desktop can expand in 2D by adding virtual desktops that extend to outside of your monitor, then expand into 3D by adding Activities in front of your monitor and you could scroll through them like the Flip Switch Task Switcher to show that only one of them is shown on your monitor. This might help some people build their mental model more easily that trying to explain the terms.

AGuiFr wrote: In System Settings, default settings are defined. In the activity manager, for each activity, you can choose to either use the default, or override the settings with something activity-specific. If a setting is overridden, a warning is displayed in the corresponding System Settings section, “this default setting is overridden in the current activity”, and links to the corresponding section in the activity manager.

I really like this idea. If the unchanged system setting is also shown in the activity-specific settings, then this also allows you to have an overview of how each activity is configured. You could clone or create an Activity and then customize it entirely in a single window.

Just a thought - the structure of the KCMs has been redefined from KDE 4.x to be more intuitive and easier to navigate, the details can be found on this page https://community.kde.org/KDE_Visual_De ... pplication.

Referring to the diagram showing the old vs the new structure, somebody please correct me if i'm wrong (it happens a lot ) but there's several places where the word Workspace is used where it doesn't mean 'all the KCMs in this category will affect the current Activity only'. For example Workspace > Startup & Shutdown > Login Manager, that KCM's not constrained to just the current activity only is it?

There's been two names suggested so far, as possible replacements for Activities, they are Environments and Workspaces. If we go for Workspaces then every new user who gives them a try then goes into System settings & sees the word Workspace used there to refer to KCMs whose effect isn't confined to just the current Workspace (the new name for Activity) there's going to be confusion and it'll make decent documentation harder to write as well i'd have thought.

IMO we either need to either:

1 - make sure the way the word Workspaces is used in System settings is consistent with what we're doing if we decide on Workspaces as the new name to avoid confusion. Maybe some of the System settings structure needs a tweak here & there, maybe Login Manager could be moved elsewhere so that it all makes sense - that could be good, users would already be aware when they see the word Workspace in System settings that any changes they make in the Workspace category will only affect this Workspace (Activity).

2 - don't choose Workspaces as the new name, to me Environments still sounds like it would make sense to all users but just avoid Workspaces.

arucard wrote:I don't think you should make a hard split between the functionalities provided by Virtual Desktops and Activities. I think the overlap in functionalities allows some flexibility on the user's part. If they need only a little bit more functionality than the hard-split Virtual Desktop would provide, they need to start using Activities which might require more effort. This could discourage the user from using that extra functionality, because it's not worth using Activities. On the other hand, if you have that functionality available in Virtual Desktops, at the point where you outgrow the needs of the Virtual Desktops it should already be worth using Activities, despite the effort required, since Activities should be able to perform the same functionalities much better and with more flexibility.

You have a point, and if we don't succeed in making activities easier to use, then I agree with you. But I think we can reduce a lot the effort of using activities for a user once he understands MM1 and MM3.
Also, I think keeping at all cost the "legacy" functionality of virtual desktops prevents us from pushing the concept further. We could add functionalities not only targetted at Berna, but at Philip (the geek) or Matt (the student in Science). For instance, in case of a multi-screen setup, you could have a VD per screen and not globally as it is now. Implementing this behaviour while keeping the current advanced features would be technically difficult, if not impossible. This is only an example and I am sure that better defining the goal of Virtual Desktop and refocussing it to only MM3 would help us come up with new ideas of functionalities enabling to better fulfil this goal. In the end, this is a question of trade-off, and for sure some users would be unhappy about the change, especially if it is perceived as a way to force them to use activities. But if it can help getting more users to use virtual desktops and activities, I think this is worthwhile.

arucard wrote:I agree that the Activities come as an additional dimension, but I don't think this needs to be changed. I think that that's the whole point of Activities, having another layer of configurability for your desktop. But it may indeed also be the reason why it's so difficult to explain clearly. Perhaps it would be useful to have a concept demo video showing a mockup of how a desktop can expand in 2D by adding virtual desktops that extend to outside of your monitor, then expand into 3D by adding Activities in front of your monitor and you could scroll through them like the Flip Switch Task Switcher to show that only one of them is shown on your monitor. This might help some people build their mental model more easily that trying to explain the terms.

GNOME, Cinnamon, ElementaryOS, Mac OS X and Windows 10 all present virtual desktops on only one dimension. Mockups of Unity 8 also show an horizontal list for displaying virtual desktops. Of course that does not mean that we have to follow the trend, but I think that if there is such a unanimity, there are good reasons. Apple and Microsoft especially have the resources to conduct extensive user testings and I don't think this choice is based only on theoretical considerations.
In Plasma, there are 2 dimensions for virtual desktops, 1 for activities, and you can eventually add an additional layer with form-factor modes (Plasma Desktop/Active/MediaCenter). You cannot expect a user as Berna to understand all these layers, they must be reduced if Berna is the main target, otherwise we consider that Virtual Desktops and Activities are features only for Philip and Matt.

I agree that the overlapping functionality should only be available as long as it doesn't interfere with improving either one of these parts. Like you said, there is a trade-off here to keep people from feeling like they're being forced to use something else. So the legacy functionality can be removed from one part as long as it's simple enough to keep using it in the other part (ideally without the user needing to reconfigure anything, so they wouldn't even notice).

AGuiFr wrote:GNOME, Cinnamon, ElementaryOS, Mac OS X and Windows 10 all present virtual desktops on only one dimension. Mockups of Unity 8 also show an horizontal list for displaying virtual desktops. Of course that does not mean that we have to follow the trend, but I think that if there is such a unanimity, there are good reasons. Apple and Microsoft especially have the resources to conduct extensive user testings and I don't think this choice is based only on theoretical considerations.
In Plasma, there are 2 dimensions for virtual desktops, 1 for activities, and you can eventually add an additional layer with form-factor modes (Plasma Desktop/Active/MediaCenter). You cannot expect a user as Berna to understand all these layers, they must be reduced if Berna is the main target, otherwise we consider that Virtual Desktops and Activities are features only for Philip and Matt.

From what I can tell, Gnome also allows a 2D layout of their Virtual Desktops though it's configured in a horizontal layout by default. In some of the later demo video's, I also noticed that they are showing their Virtual Desktops vertically when in an overview. So I don't think it's necessarily true that there's unanimity here. I didn't really look at any of the others you mentioned since, as I explain below, I think that it might not even be relevant. It should also be noted that I didn't really find any functionality similar to Activities in other desktop environments either (though I'll admit I didn't look very hard).

I think that there's a mismatch between the mental model you're suggesting and the functionality we're trying to describe. I agree that it would be much easier for users to create a mental model in the way that you describe, using a 2D space with Virtual Desktops in one direction and Activities in the other. However, it seems to me like this would also give the impression that Activities, like Virtual Desktops, are a persistent presence in your workspace. The impression would be that you can configure them, but once they're there, they stay there and anything you put on them, stays on them. This makes the Activities seem quite similar to Virtual Desktops, which is exactly the impression we're trying to avoid. While it is true that Activities can be used this way, they can also be started and stopped. This is what makes them so useful in the long run and what really sets them apart from Virtual Desktops which are just meant to give you more screen space. You can, for example, have your "Software Development" Activity configured properly, using all your machine's resources, then just stop the Activity when you want to do something else, releasing those resources until you want to resume that Activity which might be days later. This would be more accurately visualized with a 3D mental model since the user would have to understand that only one 2D plane of that 3D model can be shown on their monitor but many of them can exist, whether they are active or not. Unfortunately, this is indeed more difficult to get your head around. That's why I suggested a visual demonstration of this mental model in some kind of concept or feature demonstration video that clearly shows your desktop on your (physical) monitor, your virtual desktops in the same 2D plane stretching outside of the physical monitor, then activities that copy the entire 2D plane and show it on top of the older one. Then you can cycle through your 2D planes (Activities), select other Virtual Desktops in the same 2D plane and pause/resume some Activities by greying out the entire 2D plane (with a voice-over or on-screen text explaining what is happening as the visual representation of the mental model changes).

The distinction to be made here is between the problem of "I need more space for what I'm doing", solved by using one or more Virtual Desktops, and "I keep having to reconfigure my workspace for certain specific things I do (like developing software, performing work duties, playing video games)", currently usually solved by either configuring a Virtual Desktop for this specific thing or creating a different user to log in with or simply not solved at all, but this would be much better solved by using a different Activity for each of those specific things. So the mental model should reflect the difference between these two problems.

On a side-note, I noticed that Gnome is actually using the term "Workspace" for their virtual desktops. They also have an "Activities" overview, which seems like a Present Windows type overview of the current windows, but also includes an overview of their virtual desktops (their Workspaces) in a vertical layout. This seems very similar to how the Activities Switcher is showing different Activities in Plasma 5. This could make matters even more confusing, since it seems like we now have 2 similarly named overviews (Activities Switcher and Activities overview )of 2 entirely different things (Activities and virtual desktop, called Workspace in Gnome) that are already being confused with one another. Just thought to mention it, since it might relate to what terms should be used for these concepts in Plasma.

arucard wrote:From what I can tell, Gnome also allows a 2D layout of their Virtual Desktops though it's configured in a horizontal layout by default. In some of the later demo video's, I also noticed that they are showing their Virtual Desktops vertically when in an overview. So I don't think it's necessarily true that there's unanimity here. I didn't really look at any of the others you mentioned since, as I explain below, I think that it might not even be relevant. It should also be noted that I didn't really find any functionality similar to Activities in other desktop environments either (though I'll admit I didn't look very hard).

I did look only at the default. There might indeed be some third-party extensions which bring back a 2D desktop grid in GNOME 3. But by default, all of them show only a list. Here are some screenshots so that not everybody has to look for it : GNOME 3, Unity 8, Cinnamon, elementary OS, Mac OS X Lion, Windows 10.
I couldn’t find any feature similar to activities as well. Some environments provide a fast way to switch between user sessions if more than one are opened at once, which could cover some aspects of activities, but KDE is definitely leading on this concept.

arucard wrote:I think that there's a mismatch between the mental model you're suggesting and the functionality we're trying to describe. […] it seems to me like this would also give the impression that Activities, like Virtual Desktops, are a persistent presence in your workspace. The impression would be that you can configure them, but once they're there, they stay there and anything you put on them, stays on them. This makes the Activities seem quite similar to Virtual Desktops, which is exactly the impression we're trying to avoid.

I don’t know if you had a look at the desktop effect mockup I made in the other thread, but I think this impression might come from it. My design skills are quite limited and it can certainly be improved to avoid it. Conceptually, all that I am saying is that if we limit desktops to one spatial dimension (which most competitors are doing), we can use the other one for representing activities. I don’t think that this would necessarily reinforce the confusion between virtual desktop and activities. This would be the case if activities and desktops were always presented together: everytime you do something about activities, you also see desktops, and vice versa. I think that a desktop effect enabling to send a window to another desktop or another activity would make sense, as both actions are quite similar. IMHO, it would be better to display only started activities in this effect, leaving starting/stoping/configuring activities in the activity manager (separate for the desktop effect). As told previously, I think this activity manager should even be extended to centralise all activity-related settings in one place. But maybe merging the desktop effect and the activity manager in one single tool (clearly displaying that an activity can be started and stopped) would avoid this confusion. I prefer to keep them separate as it enables to use only desktops and not see any feature related to activities if a single activity is started.

Also, one feature which would make clear that activities are something different than virtual desktops would be to enable the user to move a whole desktop from one activity to another (maybe also through the same desktop effect). This would be possible if desktops were considered belonging to an activity and if each activity could have a different number of desktops.

Anyway, If we keep desktops as two spatial dimensions and activity as a third dimension, I can’t really figure out how we could make a desktop effect regrouping both activities and desktops which would be simple enough to use. Or are you suggesting that it is better to have one effect only for desktops, and another one only for activities ?

arucard wrote:This would be more accurately visualized with a 3D mental model since the user would have to understand that only one 2D plane of that 3D model can be shown on their monitor but many of them can exist, whether they are active or not. Unfortunately, this is indeed more difficult to get your head around. That's why I suggested a visual demonstration of this mental model in some kind of concept or feature demonstration […]

Making some tutorials is a possible solution, but I think that if Berna is the target user, the feature should be as self-explanatory as possible. I think a tutorial is good for showing a specific feature, but not for building an entire mental model (the input from an actual usability expert on this point would be valuable). This would also raise a lot of challenges: how to make sure that Berna sees this video, that she can see it again until she understands, that Matt is not bothered by it, etc… I don’t say these issues are unsolvable, but I think such tutorials should be reserved for presenting to Berna the advanced features which are developed for Philip or Matt.

AGuiFr wrote:

arucard wrote:From what I can tell, Gnome also allows a 2D layout of their Virtual Desktops though it's configured in a horizontal layout by default. In some of the later demo video's, I also noticed that they are showing their Virtual Desktops vertically when in an overview. So I don't think it's necessarily true that there's unanimity here. I didn't really look at any of the others you mentioned since, as I explain below, I think that it might not even be relevant. It should also be noted that I didn't really find any functionality similar to Activities in other desktop environments either (though I'll admit I didn't look very hard).

I did look only at the default. There might indeed be some third-party extensions which bring back a 2D desktop grid in GNOME 3. But by default, all of them show only a list. Here are some screenshots so that not everybody has to look for it : GNOME 3, Unity 8, Cinnamon, elementary OS, Mac OS X Lion, Windows 10.
I couldn’t find any feature similar to activities as well. Some environments provide a fast way to switch between user sessions if more than one are opened at once, which could cover some aspects of activities, but KDE is definitely leading on this concept.

I think Gnome actually supports a 2D desktop grid in its standard functionality, or at least it used to. But I know it's pretty easy to configure in Plasma right now, so moving to a 1D desktop layout would actually remove some degree of configurability in Plasma. I don't think virtual desktop functionality should be limited to accommodate the Activities functionality.

AGuiFr wrote:

arucard wrote:I think that there's a mismatch between the mental model you're suggesting and the functionality we're trying to describe. […] it seems to me like this would also give the impression that Activities, like Virtual Desktops, are a persistent presence in your workspace. The impression would be that you can configure them, but once they're there, they stay there and anything you put on them, stays on them. This makes the Activities seem quite similar to Virtual Desktops, which is exactly the impression we're trying to avoid.

I don’t know if you had a look at the desktop effect mockup I made in the other thread, but I think this impression might come from it. My design skills are quite limited and it can certainly be improved to avoid it.

The reason I think there's a mismatch actually doesn't have anything to do with how you describe it. I think the incorrect impression comes from the fact that if you have a 2D plane containing both activities and virtual desktops shown on a physical monitor it becomes too easy to confuse the 2 concepts because of 2 things:
1. The monitor can also display a 2D plane and it becomes natural to think you can slide your monitor across this 2D plane to view the different virtual desktops and activities. If some of them are paused or stopped, you suddenly have to skip them or move them to the edge of the 2D plane. It suddenly feels less natural to think of them in this way.
2. Having the virtual desktops and activities shown in the same 2D plane makes them seem too similar. If the only difference shown to the user is that you can move horizontally to one and vertically to the other, it would be too easy for them to think of them as the same. This could be especially problematic when 2D layout is removed from virtual desktops and people start to think they can use Activities to get it back (and then get annoyed by the fact that the desktops in their top row are acting so different from the ones in the bottom row). This is why I believe there should be a disconnect between the way virtual desktops are shown and the way activities are shown.

AGuiFr wrote:Also, one feature which would make clear that activities are something different than virtual desktops would be to enable the user to move a whole desktop from one activity to another (maybe also through the same desktop effect). This would be possible if desktops were considered belonging to an activity and if each activity could have a different number of desktops.

Anyway, If we keep desktops as two spatial dimensions and activity as a third dimension, I can’t really figure out how we could make a desktop effect regrouping both activities and desktops which would be simple enough to use. Or are you suggesting that it is better to have one effect only for desktops, and another one only for activities ?

I actually agree that it might be useful to move a whole desktop to another activity, as I agree with most of the configuration work-flows you described. As for the desktop effect, I do think that there should be one desktop effect for virtual desktops and another for activities. In fact, there might not even be a desktop effect for activities at all. I think there should be one single consistent way of moving things to other Activities, e.g. like the "Activities" option in the context menu for files where you can link it to an Activity. This makes it recognizable for the user and they can associate it with Activities instead of virtual desktops. This action might include a desktop effect or not, depending on how the action is made and what exactly it does. It doesn't seem strange to me that you don't have a desktop effect when, for example, you move a window to a different activity, since that activity is not the one you're currently working in.

AGuiFr wrote:

arucard wrote:This would be more accurately visualized with a 3D mental model since the user would have to understand that only one 2D plane of that 3D model can be shown on their monitor but many of them can exist, whether they are active or not. Unfortunately, this is indeed more difficult to get your head around. That's why I suggested a visual demonstration of this mental model in some kind of concept or feature demonstration […]

Making some tutorials is a possible solution, but I think that if Berna is the target user, the feature should be as self-explanatory as possible. I think a tutorial is good for showing a specific feature, but not for building an entire mental model (the input from an actual usability expert on this point would be valuable). This would also raise a lot of challenges: how to make sure that Berna sees this video, that she can see it again until she understands, that Matt is not bothered by it, etc… I don’t say these issues are unsolvable, but I think such tutorials should be reserved for presenting to Berna the advanced features which are developed for Philip or Matt.

I was actually not suggesting to make a tutorial on how to use virtual desktops and activities, but more of a concept demonstration. It should only show some basic representation of things and not actually show the real functionality as it's being used. This is the kind of video you often see when new products are introduced by companies and is mostly used for promotion purposes. Part of that promotion is to get the user to build a basic mental model of it so they will be more likely to try it out in the first place. KDE already makes video's like this that showcase new functionality by recording its use on a computer, usually shown on the corresponding release page for Plasma. This is similar to that but a bit more abstract since you have to show an overview of the functionality and not the functionality itself. I think this is necessary for more complex features and it's already quite common with commercial companies.

I like the idea of presenting Activities & Desktops to the user in a '2D for Desktops' & '3D for Activities' model. It would be good to have both accessible from the same UI, but at the same time keeping them separate enough to prevent confusion between the two concepts. To demonstrate the kind of thing i mean i've done a very quick & dirty mockup - warning, very rough mockups ahead!

In this mockup Desktops have been renamed Views and Activities have been renamed Environments.

We have a pretty standard screen to switch Views (aka Virtual Desktops) as normal, something like the screen below but a drop down menu has been added to it to allow access to the Environments (aka Activities) too, something like this:


When the Environments drop down is clicked it shows something like this:


If you click on an Environment in the drop down list then you're taken back to the screen of Views (the first of the two images above) but obviously the views presented on that page change to show the Views that you've setup for the Environment that you've just chosen.

I'm just trying to get across the concept, ignore the style, preview sizes etc. etc. etc. it's just the idea of having both accessible from the same page for convenience but at the same time keeping the two ideas separate enough.

To sum up our discussion so far, I think we have two main points of disagreement where a decision is absolutely needed. Please rephrase if you are not happy with my formulations:

• Should activities and virtual desktops offer features only associated with respectively MM1 and MM3, or should the current overlap be maintained in order to allow more flexibility and not to break the workflow of some of Plasma’s current users?
• Should all features related to organising windows (move to another desktop or activity) be presented in a single desktop effect, or should this be avoided (separate desktop effects or an effect only for desktops)?

We obviously have different views on both questions. I invite anyone with constructive arguments to answer these two questions.

Now answering some of your points more specifically:

arucard wrote:1. The monitor can also display a 2D plane and it becomes natural to think you can slide your monitor across this 2D plane to view the different virtual desktops and activities. If some of them are paused or stopped, you suddenly have to skip them or move them to the edge of the 2D plane. It suddenly feels less natural to think of them in this way.

Displaying desktops horizontally and activities vertically doesn’t mean this has to be a continuous 2D plane. A continuous plane is adapted to represent the idea of space extension, so it is suited for desktops. I think the transition between two activities should not be a continuous slide as it is for desktops.
Actually, if desktops are defined as belonging to an activity, each activity would have its own active desktop. Currently, if you change the active desktop in one activity, it also affects other activities, which makes little sense. So when you activate an activity, you would go back to the previously active desktop in this activity. Actually, this seems to me the most logic thing to do if a different number of desktops per activity is permitted. So if you are on activity1 desktop3 and go “down” to activity2, you would not necessarily end up on activity2 desktop3, but on the previously active desktop.
Also, skipping a stopped activity would not appear sudden to the user, as he would have stopped it before. Displaying only running activities would actually help users to understand better the effect of stopping an activity, as there would be a visual representation of a running but inactive activity and not of a stopped activity. Currently, you have to look for memory usage to understand what it does exactly.

arucard wrote:2. Having the virtual desktops and activities shown in the same 2D plane makes them seem too similar. If the only difference shown to the user is that you can move horizontally to one and vertically to the other, it would be too easy for them to think of them as the same. This could be especially problematic when 2D layout is removed from virtual desktops and people start to think they can use Activities to get it back (and then get annoyed by the fact that the desktops in their top row are acting so different from the ones in the bottom row).

Activities would act differently only if the user would have configured them differently in the first place. I don’t think this would come as a surprise to them. Some users might indeed want to use activities to get a 2D grid back, but the exact same behaviour would not be possible. With the current desktop grid, going right then down is the same as going down then right. With my proposed solution, you go right to go to the next desktop in the current activity, and when you go down, you go to the previously active desktop of the next activity. Going right then down is different from going down then right.

arucard wrote:As for the desktop effect, I do think that there should be one desktop effect for virtual desktops and another for activities.

I think that showing both on the same effect (2D or 3D) might actually help them understanding the difference between the two concepts, as they could play around with the windows, see that they can move a window both to a desktop or an activity, that they can move a desktop to an activity, that all desktops have the same wallpaper but activities have different one, etc… All centralised in a single place. If there are two separate desktop effects, they are likely to look very similar and users might focus on the similarities (I can send a window to both another desktop or another activity, what is the difference?).

arucard wrote:In fact, there might not even be a desktop effect for activities at all. I think there should be one single consistent way of moving things to other Activities, e.g. like the "Activities" option in the context menu for files where you can link it to an Activity.

A feature accessible only by right-clicking is basically non-existent for a user as Berna! The context menu is a shortcut be KDE’s Human Interface Guidelines recommends to always provide another way to access the feature than the context menu. So this should be something else than an action in the context menu.

arucard wrote:It doesn't seem strange to me that you don't have a desktop effect when, for example, you move a window to a different activity, since that activity is not the one you're currently working in.

To clarify, when I speak about “desktop effect”, I actually think about a feature similar to the current “present windows” or “desktop grid” effect, not like the animation which happens when you reduce or maximise a window. It has more to do with a functionality than an effect really.

arucard wrote:I was actually not suggesting to make a tutorial on how to use virtual desktops and activities, but more of a concept demonstration.

I think this will not be sufficient at all. For instance, computers used by the city of Munich run Plasma, and I am quite sure none of the municipal workers there will ever see a video produced by the KDE promo team. It must be directly accessible within Plasma if you want it to be seen outside the KDE community.

ken300 wrote:I like the idea of presenting Activities & Desktops to the user in a '2D for Desktops' & '3D for Activities' model. […] I'm just trying to get across the concept, ignore the style, preview sizes etc. etc. etc. it's just the idea of having both accessible from the same page for convenience but at the same time keeping the two ideas separate enough.

I am sorry, but I don’t understand how to move a window from one desktop to another? From one activity to another? These should be the primary features ; helping Berna understand the difference between activities and virtual desktops is only a secondary objective.

AGuiFr wrote:I am sorry, but I don’t understand how to move a window from one desktop to another? From one activity to another? These should be the primary features ; helping Berna understand the difference between activities and virtual desktops is only a secondary objective.

I agree, it wasn't supposed to be a fully thought out 'this is exactly how we should have it' proposal.

There's quite a few things that need addressing to overhaul Activities / Desktop to make them more accessible and this was an idea that i had to make both Activities and Desktops accessible from the same place whilst trying to maintain some kind of mental division between them so that one concept stayed quite separate from the other - that's all. It was really meant to stimulate discussion as much as anything else and for people to look at it, identify problems with it (like how do we move windows from one Desktop to the other like you said) and then between us we can come up with solutions - likely involving tearing up this idea and starting again from scratch but these mockups will have served a purpose!

It might be that when the Activities / Desktop discussion has progressed a bit & the UI to access them is being considered, someone might be thinking about 'how do we keep Desktops & Activities mentally separate' and remember that 'Ken suggested something a while ago that wasn't that bad - let's maybe use an element of that idea as a starting point & see what better ideas we can come up with' - if the only thing that is kept from my idea is to have the Activities and Desktops accessed in different ways (change the Desktops by clicking on the previews, change Activites with a different control whether it's a drop down or something else) then that's great - the mockups have done their job . The idea that i suggested might not be a solution but might lead to someone having a much better idea themselves.

Is there anything stopping us being able to drag & drop windows from Desktop to Desktop on the View screen (the first one in the post above)?

We couldn't drag & drop windows between different Activities though, but i can't think of a UI that would allow that whilst keeping Activities / Desktop mentally separate to avoid confusion and a cluttered UI - maybe someone else can