See upcoming testcase. ###!!! ASSERTION: max element width exceeded desired width: 'PR_FALSE', file ../../../mozilla/layout/tables/nsTableRowFrame.cpp, line 971 This seems to be a result of the checkin of the first patch in bug 292690. With the patch I get this assertion in my debug, without it not (tested with the bugzilla which assert appr. 5 times with the patch).

This should assert, it does for me.

I think we should just remove this assertion. Right?

I think we should fix layout to avoid to report crappy MEW values ;-), at least with the patch that I attached to bug 292690 it does not assert.

it probably should be rather a warning that makes clear that things are already going south

Robert is right. We should just remove the assert, it is bogus. The block below the cell reports a large MEW but did get during the incr. reflow only a small available width as the column is very thin (the first cell is empty which causes 0 witdh on the first colum). The table rebalances later based on the MEW requirements so there is no crappy MEW. The assertion is from a time (2001) when people (chris and I) assumed that a block should never shrink below MEW. But now that we have the table frames handling overflow the block des. size can shrink below MEW.

Comment on attachment 185979 [details] [diff] [review] patch [Checked in: Comment 7] just removes the assertion, debug-build only, so no risk

Comment on attachment 185979 [details] [diff] [review] patch [Checked in: Comment 7] mozilla/layout/tables/nsTableRowFrame.cpp 3.368