Tested 4-29-08 Win32 build under Winnt4.0J and Window 95J. Step of reproduce 1. Go to the above URL 2. Select menu View|Default Character Set->Korean(EUC_KR) Apprunner hangs. It does not do anything.

again, this is caused by illegal euc-kr. The first characters are 0xa0a0, which is not legal euc-kr. Euc-kr start w/ 0xa1a1

this is cause by the ignorance of error handling in nsScanner::Append method. I have the fix but I preferred we check in after M5. I think we can down grade this to M6 because this will only hang when the first byte of the html have error in converter, which I think is a rare case. The fix is as below, need rickg to review and approve for M6 check in.... Z:\mozilla\htmlparser\src>cvs diff -c nsScanner.cpp Index: nsScanner.cpp =================================================================== RCS file: /m/pub/mozilla/htmlparser/src/nsScanner.cpp,v retrieving revision 3.41 diff -c -r3.41 nsScanner.cpp *** nsScanner.cpp 1999/04/21 04:02:14 3.41 --- nsScanner.cpp 1999/05/01 00:51:46 *************** *** 244,257 **** PRBool nsScanner::Append(const char* aBuffer, PRUint32 aLen){ if(mUnicodeDecoder) { ! PRInt32 unicharLength = 0; ! PRInt32 srcLength = aLen; ! mUnicodeDecoder->Length(aBuffer, 0, aLen, &unicharLength); ! PRUnichar *unichars = new PRUnichar [ unicharLength ]; ! nsresult res = mUnicodeDecoder->Convert(unichars, 0, &unicharLength,aBuffe r, 0, &srcLength ); ! mBuffer.Append(unichars, unicharLength); ! delete[] unichars; ! mTotalRead += unicharLength; } else { --- 244,272 ---- PRBool nsScanner::Append(const char* aBuffer, PRUint32 aLen){ if(mUnicodeDecoder) { ! PRInt32 unicharBufLen = 0; ! mUnicodeDecoder->Length(aBuffer, 0, aLen, &unicharBufLen); ! PRUnichar *unichars = new PRUnichar [ unicharBufLen ]; ! nsresult res; ! do { ! PRInt32 srcLength = aLen; ! PRInt32 unicharLength = unicharBufLen; ! res = mUnicodeDecoder->Convert(unichars, 0, &unicharLength,aBuff er, 0, &srcLength ); ! mBuffer.Append(unichars, unicharLength); ! mTotalRead += unicharLength; ! // if we failed, we consume one byte by replace it with U+FFFD ! // and try conversion again. ! if(NS_FAILED(res)) { ! mBuffer.Append( (PRUnichar)0xFFFD); ! mTotalRead++; ! aBuffer += srcLength + 1; ! aLen -= srcLength + 1; ! } ! } while (NS_FAILED(res) && (aLen > 0)); ! // we continue convert the bytes data into Unicode ! // if we have conversion error and we have more data. ! ! delete[] unichars; } else {

Change summary from "Changing Character Set on no Meta Euc-KR page causes HANG" to "illegal bytes in the first byte of docuemnt cause hang in double byte charset"

cc'd rickg

moving to m6.

Same thing happens as follows 1. Go to http://babel/tests/browser/bft/euc-jp/bft_browser_link.euc-jp.html 2. Select menu View|Default Character set->UTF-8 Apprunner hangs. Once I see Frank's fix, I will test this, too.

fix and check in.

I verified this in 5-17-09 Win32 build.

Moving all Apprunner bugs past and present to Other component temporarily whilst don and I set correct component. Apprunner component will be deleted/retired shortly.