386449 - add assertions for invariants of primary frame relationship

Status: REOPENED

(Core :: Layout, defect)

Description

[David Baron :dbaron: (⌚️UTC-4, no longer working on Mozilla)]

We should add assertions to enforce invariants of the primary frame map, as follows:

 * nsFrameManager::SetPrimaryFrameFor should have the assertion (currently in #if 0) that aPrimaryFrame->GetContent() == aContent enabled

 * nsFrameManager::SetPrimaryFrameFor should have the warning that entry->frame is null (currently #ifdef DEBUG_dbaron) enabled for everybody and turned into an assertion

 * nsFrameManager::RemoveAsPrimaryFrame should assert that entry->frame == aPrimaryFrame rather than checking it

Comment 1

[Jesse Ruderman]

The primary frame map was removed in bug 500882.

Comment 2

[Jesse Ruderman]

Or are some of these assertions still relevant to the "primary frame mapping" that now exists as a pointer held by the frame?

Comment 3

[David Baron :dbaron: (⌚️UTC-4, no longer working on Mozilla)]

These are still relevant to primary frame pointer in the content node.

Comment 4

[Boris Zbarsky [:bzbarsky]]

> * nsFrameManager::SetPrimaryFrameFor should have the assertion (currently in
> #if 0) that aPrimaryFrame->GetContent() == aContent enabled

This should move (still in #if 0 pending the <area> mess being fixed) into nsIContent::SetPrimaryFrame, right?

> * nsFrameManager::SetPrimaryFrameFor should have the warning that entry->frame
> is null (currently #ifdef DEBUG_dbaron) enabled for everybody and turned into
> an assertion

This was effectively done in bug 536716, I think.

> nsFrameManager::RemoveAsPrimaryFrame should assert that entry->frame ==
> aPrimaryFrame rather than checking it

This part I'm not sure has a new-world equivalent, nor that it should.