> On Aug. 17, 2014, 2:36 p.m., Eike Hein wrote:
> > Looks good, but: I'm aware of at least one downstream that would like to 
> > hide activities support from plasmashell for complexity reasons. Is there a 
> > way to disable activities support, and does it have an API we could use to 
> > conditionally hide the option in Folder?
> 
> Ivan Čukić wrote:
>     There is no way to disable activities. At least, not a supported way.
>     
>     I *do* sometimes test a few things without the activity manager running, 
> but it was decided at one point (a long time ago) that we will not be 
> supporting that setup in any manner.
>     
>     The only component that is always guaranteed to work without activities 
> is kwin, because it needs to work with other environments beside Plasma.
>     
>     From my point of view, if a distribution wants something like that, they 
> need to be prepared to maintain their own patches which I don't think is very 
> manageable in the long-run.
>     
>     
>     * interestingly enough, there still is a hidden checkbox in the 
> activities kcm for disabling the system, but it was voted out as soon as I 
> created it. It has absolutely no effect now.
> 
> Eike Hein wrote:
>     > From my point of view, if a distribution wants something like that, 
> they need to be prepared to maintain their own patches which I don't think is 
> very manageable in the long-run.
>     
>     I agree, having them fork the code downstream all over the place is 
> pretty ugly. So the only recourse is to leave this patch out for now then.
> 
> Ivan Čukić wrote:
>     The problem with leaving the patch out is that it is a regression 
> compared to 4.1x (pre-baloo).
>     
>     And if we do it like this, no new feature (at least, activity-related) 
> will ever get in.
> 
> Eike Hein wrote:
>     I actually feel really, really uncomfortable taking a firm stance here, 
> because I can see both sides.
>     
>     - Making Activities optional can inhibit its uptake, preventing the 
> implementation of functionality that relies on them being there, and so 
> possibly functionality that justifies their existence. This is slightly 
> similar to the old "Nepomuk on by default" debate.
>     
>     - OTOH Plasma does claim to care about being a framework for downstreams 
> to customize or implement new interfaces, and requiring forking+patching 
> isn't supporting that cleanly. That some downstreams may want to offer 
> interfaces that don't use Activities is legitimate. There's also value in the 
> "if downstreams are going to do it anyway, we might as well maintain it 
> upstream" stance that's popular through much FOSS. We've also intentionally 
> modularized and decoupled a lot of things before as good engineering practice.
>     
>     The only compromise I can come up with is not to make the "Disable 
> activities" knob visible in the UI, but make it a ISV-level config option 
> somewhere, or even a build-time option. A precedent: Some downstreams don't 
> like the Desktop Toolbox. In Plasma 5 we provide a clean way to hide it, you 
> can just put the .desktop file for the package in your distro profile overlay 
> and add Hidden=true. Something like this for Activities might be a good idea 
> - perhaps as simple as not autostarting the daemon and having the UI adapt to 
> it.
>     
>     Of course we could also put an ISV-level option into Folder, i.e. have a 
> UseActivities config key distros can set to false via desktop scripting and 
> hide the option. But this has the downside of reimplementing it in every 
> widget instead of having something central.
> 
> Marco Martin wrote:
>     note that not using activities is different to not having the daemon 
> running or having the linking database active.
>     not offering an activity switcher is one thing not having the daemon 
> running will break things, no ways around it.
>     If is really needed to not have it, the config ui can check if the daemon 
> is not running not show the option, that may be a solution.
>     "expect bug" is a constant tough
> 
> Marco Martin wrote:
>     Another option is to put an option in defaults of the shell package (or 
> look&feel?) on wether those options should be visible or not
> 
> Eike Hein wrote:
>     Marco: Hmm, that sounds cool. How would that look like from a code POV, 
> i.e. how exactly would Folder adapt to the LNF option?
> 
> Marco Martin wrote:
>     would have to be accessed by the c++ part..
>     that may mean also that i'll finally be forced to make the 
> lookandfeelaccess class finally available in a library somewhere (what that 
> class does is exposing a filePath() method to take files from the l&f package 
> performing a fallback to the default one when the configured l&f package 
> doesn't provide a certain file).
>     As first thing there may be just an hidden option.. could also be cool 
> having a set of default configs for applets (as soon as created) either in 
> the shell or l&f package, I'm not too sure how i would do that tough
> 
> Eike Hein wrote:
>     I guess C++ would be OK for me; Folder already has a "SystemSettings" 
> class in the plugin for things like double click vs. single click and double 
> click interval that I could add a prop to. So this may be a way forward.
> 
> Ivan Čukić wrote:
>     It would be nice if this was accessible directly from qml - so that all 
> applets can access it in the same way.
> 
> Marco Martin wrote:
>     "It would be nice if this was accessible directly from qml - so that all 
> applets can access it in the same way."
>     there may be several apis for that.
>     maybe having just initializations of the settings?
>     like: there is a file in the package, "plasmoiddefaults", is a config 
> file:
>     [org.kde.plasma.folder]
>     showActivitiesOption=false
>     [whateverotherplasmoid]
>     ...
>     
>     the initial plasmoid.configuration object will be initialized with the 
> value there instead of the actual default in its xml...
>     
>     how does that sound?

^ Hmm, doesn't that defeat the purpose of centralizing the option by explicitly 
enumerating plasmoids again? What if there's a new plasmoid, or a third-party 
one? You end up having to reissue the LNF.


- Eike


-----------------------------------------------------------
This is an automatically generated e-mail. To reply, visit:
https://git.reviewboard.kde.org/r/119813/#review64674
-----------------------------------------------------------


On Aug. 17, 2014, 2:32 p.m., Ivan Čukić wrote:
> 
> -----------------------------------------------------------
> This is an automatically generated e-mail. To reply, visit:
> https://git.reviewboard.kde.org/r/119813/
> -----------------------------------------------------------
> 
> (Updated Aug. 17, 2014, 2:32 p.m.)
> 
> 
> Review request for Plasma, Eike Hein, Marco Martin, and Sebastian Kügler.
> 
> 
> Repository: plasma-desktop
> 
> 
> Description
> -------
> 
> This is port of the feature that the folder view had before baloo replaced 
> nepomuk - the configuration option to show the files that are linked to the 
> current activity (activities:/current/).
> 
> It requires the latest master of kactivities which ship with the new kio 
> implementation.
> 
> 
> Diffs
> -----
> 
>   containments/folder/package/contents/ui/ConfigLocation.qml dc97e81 
> 
> Diff: https://git.reviewboard.kde.org/r/119813/diff/
> 
> 
> Testing
> -------
> 
> 
> Thanks,
> 
> Ivan Čukić
> 
>

_______________________________________________
Plasma-devel mailing list
Plasma-devel@kde.org
https://mail.kde.org/mailman/listinfo/plasma-devel

Reply via email to