Comment on attachment 155165 [details] [diff] [review] patch >Index: content/base/src/nsDocumentEncoder.cpp >+ AppendUTF16toUTF8(mMimeType, progId); AppendASCIItoUTF16 would work too, no? Assuming that's actually a MIME type.. r+sr=bzbarsky either way.

I'll do the second of the nsPlainTextSerializer functions in Bug 254298, not here, since it's not currently guaranteed that it's ascii

I'm using ToUTF8 to avoid assertions when people put nonascii mimetypes in their html docs. checked in.

Comment on attachment 155165 [details] [diff] [review] patch This patch here regressed bug 299529, and propbably other bidi cases as well: > nsIFrame *blockFrame = currentFrame; > nsIFrame *thisBlock = nsnull; > PRInt32 thisLine; >- nsCOMPtr<nsILineIteratorNavigator> it; >+ nsILineIteratorNavigator* it; // This is qi'd off a frame, and those aren't >+ // refcounted > result = NS_ERROR_FAILURE; > while (NS_FAILED(result) && blockFrame) > { > thisBlock = blockFrame; > blockFrame = blockFrame->GetParent(); > if (blockFrame) { >- it = do_QueryInterface(blockFrame, &result); >+ CallQueryInterface(blockFrame, &it); > } > } > if (!blockFrame || !it) > return NS_ERROR_FAILURE; Notice that with the patch, |result| is never assigned to inside the loop. As a result, the loop breaks only when blockFrame is null - which causes the method to abort with NS_ERROR_FAILURE immediately afterwards. This renders nsSelection::GetPrevNextBidiLevels() useless.

Reopening to get regression fixed (I hope this is proper procedure).

Actually, you should file new bugs on regressions, in general.

Comment on attachment 188858 [details] [diff] [review] regression patch fixes a regression with bidi cursor navigation

Comment on attachment 188858 [details] [diff] [review] regression patch 1.8b3 is behind us. Requesting approval1.8b4.

regression patch checked in Checking in layout/generic/nsSelection.cpp; /cvsroot/mozilla/layout/generic/nsSelection.cpp,v <-- nsSelection.cpp new revision: 3.197; previous revision: 3.196 done