* The basic idea here is to have a reasonable middleground between
ResizeToContents and Stretch. We want to show at *least* enough for
the contents, but the remaining space should be shared between the
columns according to some proportions.
* That way you don't end up with one huge column and several tiny ones
that are just big enough but no more, but all data is still visible.
* Add scroll areas for the larger tabs and allow pipe flowchart to
resize to its minimum size if needed.
* The idea here isn't that anyone will seriously try to use the pipe
view at 500x200 but more that the minimum size doesn't become a
problem while moving around windows or panels, or resizing temporarily
* This prevents a feedback loop where the label resizes wider to make
room for the margins, then the margins get bigger to keep the image
centred/scaled properly, etc.
* This way it can be used from Qt or any other UI as well.
* The pipes are created internally and just passed as stdin to the
renderdoccmd processes instead of being named pipes.
* These structs were no longer used since the memory wholeMemBuf was
added, and they had some confusing enum usage that triggered static
analysis warnings.
* When descriptor sets are invalidated and we have stale bindings around
we cannot bind them or it will lead to invalid behaviour and crashes.
* However it's possible for set 0 to be invalid/stale and set 1 to be
valid and required, so we can't bail out at the first invalid set. We
keep iterating and bind all valid compatible sets.
* When we do a partial replay we truncate the render pass to only have
one subpass, but pipelines are created for a particular subpass. So
at creation time we create a new pipeline that points to subpass 0 so
we can use it later together with the truncated render passes.
* This is far more likely when you start capturing at the start of the
program, but it can happen any other time if you get unlucky.
* Normally we expect the preparation of initial states for programs to
get all the data ready, but if the program is created during the frame
capture then that's after preparation, so we need to instead fetch and
serialise the data at the end of the frame while serialising initial
contents.
* This isn't reliable in GL since context creation is such a mess, but
we go from the first attribs-based context create to the next present.
* Should allow capturing early setup and creation before the first
render.
* The proxy replay driver might not report an 'internal' ID that doesn't
have a true live ID, so instead we take responsibility for reporting
it. The proxy will still return its details when prompted.
baldurk