evaluate performance of ElementBoxWidget

[Ichthyostega]

ElementBoxWidget is a building block of the Lumiera UI, and expected to be used pervasively. Most notably, the display of individual media Clips within the timeline is based on this custom widget — and we can expect a typical feature film sequence to hold hundreds of clips, which must response to focus and zoom scale changes without notable delay. Moreover, representing all those clips — even just in hidden state — might turn out as a significant factor for the UI's memory consumption (and preferably we should use available memory for display buffers and caches anyway)

Topics for investigation

Setup

build a reproducible Test session populated with an excessive amount of media clips

Memory consumption

provide instrumentation to watch the memory footprint of ElementBoxWidget instances

• determine the impact of storing the full-rendered size of the IDLabel (as is done in the current implementation)
• determine the impact of memorising the full unabridged text of the label within the widget (in case of shortening the label display)
• determine the impact of caching the previous size constraint for short-circuit evaluation

Runtime

• measure the time require to hide/show all widgets on extreme zoom scale change
• measure the time of a focus change, with/without caching redundant previous constraint checks
• measure the time of an average focus change with many unconstrained widgets without caching the full-rendered size

Acceptance Criteria

• the implementation can be considered to be solid even under extreme conditions
• measurements and reasoning have been documented

[Ichthyostega]

Introduced a logical cycle with this ticket mistakenly...

It is true that we do need a roughly complete Implementation in order to get sensible performance measurements, yet the shortening of the ID Label (#1242) can only be implemented without guesswork, when such a performance survey has already been done. And without the flexible handling of ID Label, the ElementBoxWidget (#1185) can certainly not be called complete...

⟹ the most sensible thing to do is to slightly lower the requirements for the investigation,
but at least require a half way complete implementation, with relevant parts already filled in....

• lowering to the base implementations of ElementBoxWidget and Clip implementation
• but adding also GUI timeline layout structure, which implies a partial integration,
  featuring the foundation for Timeline Layout management
• and also require the draft for Zoom handling

[Undercover Agent]

Migration MasterTickets ⟼ Subtickets-plugin