It should allow make the table test (bug 570995) faster in two times because of reduction of nsOuterTableFrame::GetRowAndColumnByIndex() calls number which is bottle neck now.

try server build - http://ftp.mozilla.org/pub/mozilla.org/firefox/tryserver-builds/surkov.alexander@gmail.com-9ac34b5c7b36/

Comment on attachment 454852 [details] [diff] [review] patch r=me for the tests and the functionality. This improves perf of the table mutation test from the 35400/35350 MS of try-server build of bug 575576 to 24409/24386 MS with this try-server build. Additionally, initial loading time of the page with NVDA is vastly improved.

Comment on attachment 454852 [details] [diff] [review] patch r=me! Only nits: >+++ b/accessible/src/msaa/CAccessibleTable.cpp I'm not sure the cosmetic changes are worth the loss in history/annotation -- but I'll leave it up to you to keep them or not. >+nsXULTreeGridAccessible::GetRowAndColumnIndicesAt(PRInt32 aCellIndex, >+ PRInt32* aRowIndex, >+ PRInt32* aColumnIndex) >+{ >+ NS_ENSURE_ARG_POINTER(aRowIndex); >+ *aRowIndex = -1; >+ NS_ENSURE_ARG_POINTER(aColumnIndex); >+ *aColumnIndex = -1; Nit: this is fine or you might prefer: NS_ENSURE_ARG_POINTER(aRowIndex); NS_ENSURE_ARG_POINTER(aColumnIndex); *aRowIndex = *aColumnIndex = -1; >+ ok(false, id + ": can't get row and column indexes for cell index " + idx + "," + e); Nit: "indices"

(In reply to comment #4) > (From update of attachment 454852 [details] [diff] [review]) > r=me! Only nits: > > >+++ b/accessible/src/msaa/CAccessibleTable.cpp > > I'm not sure the cosmetic changes are worth the loss in history/annotation -- > but I'll leave it up to you to keep them or not. That's always a choice: nicer code vs cleaner history. I don't have an answer. Here I would prefer to keep. > NS_ENSURE_ARG_POINTER(aRowIndex); > NS_ENSURE_ARG_POINTER(aColumnIndex); > *aRowIndex = *aColumnIndex = -1; this shorter and nicer but not correct though nobody cares in the end I guess.

landed on 2.0 (1.9.3) - http://hg.mozilla.org/mozilla-central/rev/5425902639a5