Note that this is a double-edged sword and that docks aren't just kept in memory; the views actually continue running as normal while the dock is "closed" (closed is not closed, it's actually just moved out of view). Be careful using this with resource-heavy views and components like charts and alarm status tables as these can significantly slow down your pages if you have all of these within docks.
I have a qr scanning perspective component provided by a 3rd-party module and had this in a dock so that the operators didn't need to wait for the camera to initalise every time which was taking a few seconds. The caveat however was that they were always able to scan QR codes as the dock it was in ran regardless of being visible or not. I had to put code in place to enable scanning only when visible
True, but you can also just stop those from running when the dock is closed with some scripting. Ultimately, you’ll have to play with whatever works best for your application.
For me, there are 2 main differences that sets perspective apart from vision.
The ability to do scripts transforms on every binding. This makes it really easy to take an object oriented approach to sending data. The ability for Custom properties/parameters to be dynamic is in my opinion really what sets perspective apart from vision.
Components/Graphics do not behave nicely in vision. Scaling, Rotating, Centering, "Lining up Pipes and other graphics" can sometimes be a challenge, especially when different screen sizes are involved. Adjusting the size of the moving analog indicator is especially infuriating when trying to line it up with other components.
These are both things that can probably be corrected via software updates to vision, but currently from a design perspective, if we follow good design practices, I will recommend Perspective every time for our customers as it simply cuts down on design time with its flexibility.