XDG_CURRENT_DESKTOP

[Michael Terry]

On 07/19/2011 09:12 AM, Matthias Clasen wrote:
> I have said this repeatedly, I will say it again:
>
> I don't think such 'desktop marking' is making anything better, really.
>
> It only makes the mess bigger by proliferating wierd and unmaintainable
> desktop-specific workarounds in many places.
>
> Applications should ideally not change their behavior depending on such
> markers. If they need to use services which may or may not be present,
> check for the proper dbus service names on the bus.

I agree in the general sense that features, not environments, should be 
tested for.

But in the case of desktop file parsing, XDG defines OnlyShowIn values, 
but never defines a way to know what value to use.  I think that lack of 
definition is a pretty clear use case for such a variable.

This isn't about defining a shortcut to avoid testing for features.  
Maybe if a mini-spec gets written, it could include a section like:

"It is not recommended to assume features of a given desktop are present 
after examining XDG_CURRENT_DESKTOP.  It's best to test for features as 
you need them.  XDG_CURRENT_DESKTOP is only recommended for checking 
OnlyShowIn and NotShowIn values."

I think it still has some applications outside of that narrow scope.  
xdg-utils for example, really just wants to know the desktop 
environment; there's no feature it can test for there.

-mt