waterson: mozilla bugs & test cases

home | <

Herein lie fun HTML and XUL test cases that illustrate bugs and stuff. It's not quite as clever or well organized as dbaron's CSS playground or Hixie's tests, but it's a start. And relevant to my little corner of Mozilla fun.

Block-and-line

Incremental Reflow

While working on the reflow tree, dbaron pointed out that there could be cases where an incremental reflow targeted at a container could mask incremental reflows targeted at one of the container's children, or might not properly be propagated to the container's children.

• This test case (or a minimal version of it) demonstrate a case where an incremental dirty reflow targeted at an outer <div> could potentially mask an incremental style-changed reflow targeted at a nested frame.
• This test case (or a minimal version of it) demonstrate a case where an incremental style changed reflow targeted at an outer <div> might not properly be propagated to its children.

This revision of the work actually has the latter problem.

Floaters

Incremental Reflow

While working on eliminating nsBlockFrame::RememberFloaterDamage, I thought it would be good to come up with a couple of test cases that stress incremental floater reflow. These all trigger incremental style-changed reflows, and are self-animating:

• Left float in a sibling block.
• Right float in a sibling block.
• Left float in the parent block.
• Right float in the parent block.
• Left float with zero-sum margin adjustment.

These trigger incremental dirty reflows when you click on the document body:

• Left float.
• Right float.
• Left float, separate block.
• Right float, separate block.

There's also:

• The original test case that dbaron concocted in response to the incremental reflow tree work.
• A testcase the illustrates why we must not just test for a change in the float's width, but also for a change in its position.

Tables

Incremental Reflow

While working on the incremental reflow tree, I broke some table incremental reflow. This is a test that triggers two dirty reflows, targeted at two different rows, when you click on the page body somewhere. (It's interesting in the context of the reflow tree, because the reflow tree coalesces them, and if you've botched the iteration, you'll drop one on the floor.)

Here are some test cases for <caption> elements, which I also broke with the reflow tree landing.

• Incremental reflow targeted at caption's child. Click the page to trigger the change.
• Incremental reflow targeted at caption. Click the page to trigger the change.

Comboboxes

Incremental Reflow

More incremental reflow tree fun. This is a test case that modifies the contents of the combobox dynamically: I created it to see just exactly what the special-case code for handling the drop-down frame in nsComboboxControlFrame was doing. Apparently, it isn't doing much of anything: in theory, comboboxes shouldn't update with the incremental reflow tree changes; however, they seem to work just fine.

rods gave me three other test cases to try that are similar:

• Test #1.
• Test #2.
• Test #3.
• Test #4, or a minimized version that broke the reflow tree patch for a while.

XUL

Templates

This is a test case for a remote datasource bug (which, unfortunately, I couldn't reproduce). Both the XUL and the RDF are loaded via the http: protocol.

Signed Scripts

This is a test case for a bug that causes DOM properties to be inaccessible from a signed script. There is a follow-on bug that deals with the security manager getting confused by an XML document being created from script: specifically, it doesn't think that the new document has a cert that is compatible with the original script's principal.

You may need to install this cert and give it “software signing” privileges to actually use the test cases above.