Copying a venkman error [can't set breakpoint on inspector line because it has no code] to nc4 to report a bug about venkman (while trying to use venkman to report a bug about inspector). vnk: setCurrentSourceProvider: [object Object] ###!!! ASSERTION: Mixed wrapping data and nonwrapping data on the same line: 'mCurrentLine.Length() == 0', file f:\build\mozilla\content\base\src\nsPlainTextSerializer.cpp, line 1467 NTDLL! 77fa018c() nsDebug::Assertion(const char * 0x02000568, const char * 0x0200054c, const char * 0x02000510, int 1467) line 290 + 13 bytes nsDebug::WarnIfFalse(const char * 0x02000568, const char * 0x0200054c, const char * 0x02000510, int 1467) line 396 + 21 bytes nsPlainTextSerializer::Write(const nsAString & {...}) line 1467 + 51 bytes nsPlainTextSerializer::DoAddLeaf(int 118, const nsAString & {...}) line 956 nsPlainTextSerializer::AddLeaf(nsPlainTextSerializer * const 0x075fda24, const nsIParserNode & {...}) line 429 CNavDTD::AddLeaf(const nsIParserNode * 0x059de0c0) line 3802 + 25 bytes CNavDTD::HandleEntityToken(CToken * 0x05a6ba30) line 2168 + 18 bytes CNavDTD::HandleToken(CNavDTD * const 0x075fd1d0, CToken * 0x05a6ba30, nsIParser * 0x075fa0d0) line 906 + 12 bytes CNavDTD::BuildModel(CNavDTD * const 0x075fd1d0, nsIParser * 0x075fa0d0, nsITokenizer * 0x075fd150, nsITokenObserver * 0x00000000, nsIContentSink * 0x075fda24) line 525 + 20 bytes nsParser::BuildModel() line 1981 + 34 bytes nsParser::ResumeParse(int 0, int 0, int 0) line 1847 + 11 bytes nsParser::Parse(const nsAString & {...}, void * 0x00000000, const nsAString & {...}, int 0, int 1, nsDTDMode eDTDMode_autodetect) line 1730 + 17 bytes nsHTMLFormatConverter::ConvertFromHTMLToUnicode(nsHTMLFormatConverter * const 0x075f8100, const nsAutoString & {...}, nsAutoString & {...}) line 321 nsHTMLFormatConverter::Convert(nsHTMLFormatConverter * const 0x075f8100, const char * 0x075fb71c, nsISupports * 0x075f80c0, unsigned int 610, const char * 0x013fc8e8, nsISupports * * 0x0012fbfc, unsigned int * 0x0012fc04) line 253 + 26 bytes nsTransferable::GetTransferData(nsTransferable * const 0x075fad40, const char * 0x013fc8e8, nsISupports * * 0x0012fbfc, unsigned int * 0x0012fc04) line 383 nsDataObj::GetText(const nsACString & {...}, tagFORMATETC & {...}, tagSTGMEDIUM & {...}) line 612 nsDataObj::GetData(nsDataObj * const 0x075fb580, tagFORMATETC * 0x0012fcc0, tagSTGMEDIUM * 0x0012fc84) line 196 + 23 bytes OLE32! 77a8d301() OLE32! 77a8d1d5() OLE32! 77a5693d() USER32! 77e12e98() USER32! 77e139a3() USER32! 77e1395f() NTDLL! 77fa032f() USER32! 77e1569d() nsAppShell::Run(nsAppShell * const 0x004b6fa0) line 105 + 24 bytes nsAppShellService::Run(nsAppShellService * const 0x004b3380) line 303 main1(int 1, char * * 0x00444390, nsISupports * 0x00000000) line 1264 + 32 bytes main(int 1, char * * 0x00444390) line 1594 + 37 bytes mainCRTStartup() line 338 + 17 bytes KERNEL32! 77e97d08()

What is the data that is converted to plain text? This is because some (potentially) wrapped text has been inserted but then something else comes along, a <pre> for instance, and inserts text that can't wrap. Therefore a <pre> must flush the text buffer before text is inserted. <pre> does this right now so it don't think it's a <pre> that's caused this to happen. Hmm... There is also a "aString.First() == PRUnichar('>')" case that I don't recall having seen before. It seems as a line starting with a ">" inside a span triggers <pre>-formatted output. That must be wrong. Is that what happens for you? ("line" = chunk of text sent to the converter)

Ok. I seem to have reviewed that code (it's a fix for bug 69638) back in July. I wonder if that's the problem. Anyway, I have to see tha data that nsPlainTextSerializer is fed with to be able to say anything remotely wise.

Did this appear a couple of days ago? In that case it's caused by the checkin for bug 92331 - plain text reply indenting isn't ending lines correctly with values other than 78 chars. Adding jfrancis to the CC.

I've been a negligent QA so i have no idea when this started, Clipboard that can trigger it (as heard by nc4editbox): 'de at line 196 <chrome://inspector/content/ViewerRegistry.js> contains no code at line 207 <chrome://inspector/content/ViewerRegistry.js> contains no code at line 207 Exceptions will now stop the debug target.' Pasting the code into a blank mozilla composer page results in: <table border="0" cellpadding="0" cellspacing="0" id="output-table"> <tbody id="output-tbody"> <tr class="msg" msg-type="ERROR"> <td class="msg-data" msg-type="ERROR"><span>de at line 196<br> </span></td> </tr> <tr class="msg" msg-type="ERROR"> <td class="msg-data" msg-type="ERROR"><span>&lt;<a class="venkman-link" target="_content" href="chrome://inspector/content/ViewerRegistry.js"><span><img class="spacer-node"> chrome://inspector/<img class="spacer-node"> content/ViewerRegistry.<img class="spacer-node"> js<img class="spacer-node"> </span></a>&gt; contains no code at line 207<br> </span></td> </tr> <tr class="msg" msg-type="ERROR"> <td class="msg-data" msg-type="ERROR"><span>&lt;<a class="venkman-link" target="_content" href="chrome://inspector/content/ViewerRegistry.js"><span><img class="spacer-node"> chrome://inspector/<img class="spacer-node"> content/ViewerRegistry.<img class="spacer-node"> js<img class="spacer-node"> </span></a>&gt; contains no code at line 207<br> </span></td> </tr> <tr class="msg" msg-type="INFO"> <td class="msg-data" msg-type="INFO"><span>Exceptions will now stop the debug target.</span></td> </tr> </tbody> </table>

Yes, that's it. Study the line: </span></a>&gt; contains no code at line 207<br> It's inside a span (it was inside two span:s before the inner one ended) and the entity &gt; will be translated to the string ">" and sent to write where it gets treated as preformatted text and causes the assert. Akkana, do you remember the reasoning behind the lines: if ((mPreFormatted && !mWrapColumn) || IsInPre() --> || (!mQuotesPreformatted && mSpanLevel > 0 --> //&& Substring(aString, 0, 1) == NS_LITERAL_STRING(">"))) { --> && aString.First() == PRUnichar('>'))) { The QuotesPreformatted prefs was true until last week so I don't think this code has been active (normally). AFAIK the case can be removed, but as it was part of your fix for bug 69638, I guess it's needed. In that case we need to make the test better to only be triggered when it should be triggered.

Sorry for the delay. I believe Joe has flipped the switch to turn off mQuotesPreformatted, so the code is active now; or if it isn't, it will be soon. We do still need it.

Akkana speak truth with non-forked tonuge!

tongue? (don't worry i'm sure i make more spelling errors than you do) -- now where's that spellchecker [bug/rfe] for submittable form fields/.

i dont think anyone makes more spelling errors than me. it's cuz i look at the keyboard while typing, rather than the screen.