This is the height version of bug 367442. We have some places that set computed heights too, and I want to add another one in bug 210094. Should just make this a method.

Oh, the only interesting changes are in nsHTMLReflowState.{h,cpp}, nsComboboxControlFrame.cpp, nsFieldSetFrame.cpp, and nsTableRowFrame.cpp

Comment on attachment 272231 [details] [diff] [review] Fix Not sure why you thought the combobox and fieldset changes were interesting. In nsFieldSetFrame, it looks like you wrote SetComputedWidth instead of SetComputedHeight. In nsTableRowFrame, it looks like you dropped the PR_MAX with 0. Was that intentional? It seems wrong. In nsTableCellFrame, mind changing the casts to const_cast? r+sr=dbaron with those things fixed. Sorry for the delay.

> Not sure why you thought the combobox and fieldset changes were interesting. They were more than a direct search/replace. Only relatively interesting... > In nsFieldSetFrame, it looks like you wrote SetComputedWidth instead Yikes. Good catch. > In nsTableRowFrame, it looks like you dropped the PR_MAX with 0. Again, good catch. That wasn't supposed to happen. > In nsTableCellFrame, mind changing the casts to const_cast? Sure thing.

Comment on attachment 272231 [details] [diff] [review] Fix This replaces direct access to the mComputedHeight member of nsHTMLReflowState with a getter/setter pair. The setter makes sure to reinitialize the mVResize flag as needed, so that callers don't optimize away reflows that actually need to happen.

Comment on attachment 272231 [details] [diff] [review] Fix a181=dbaron

er, a19=dbaron. At least I checked the right box.

Checked in with fixes per comment 2.