Reproduce: Create a msg with the HTML editor with the following content and send it in "Both plain text and HTML": *hello* b@b.com Actual result: The plain text part reads: **hello** b@b.com <"b@b.com"> (ignore the quotes, that's bug #29211) Expected result: The plain text part reads: *hello* b@b.com Proposed fix: I guess, we first run the mail through the HTML enhancer mozTXTToHTMLConv::ScanHTML and then convert the result to plain text. We should convert the the result of the editor directly to plain text.

reassign to rhp

I've tried this in all sorts of combinations and I can't recreate this problem on Win32 or Linux so I am marking as such. - rhp

rhp, see <news://news.mozilla.org/38B5EC8F.9060802@netscape.com> ("Mozilla source tree is open... but..." in .seamonkey) for an example, look out for "VERY". I tried it several times before filing that bug.

Maybe I'm missing something, but please tell me what is wrong with the two images that I have attached. The first is 4.x and the second is 5.x...they look identical to me? - rhp

The problem is in the plain text part. Mozilla and 4.x display only the HTML part, if available, in the msg view, so you can see the problem only in their "View Source". You will find a "**VERY**" there.

Ok, the text enhancement was being applied to the conversion of the HTML to plain text on the sending operation. This shouldn't be done since we should be sending the original "plain text" (not what we render to the user). I have a fix in my tree and will checkin with everything else when the tree opens for M15. - rhp

rhp, I don't understand you fully. What IMO should be done (and what 4.x basically did) is the following: The HTML source (what the user generated using the HTML editor) should be run through ScanHTML and the output of this function send as HTML part. The HTML source again (HTML part of the msg) should be run through nsHTMLToTXTSinkStream (with OutputFormatted) and the output of this class send as plain text part. Guessed old code: +------------------------------> HTML part HTML source ---> ScanHTML ---+ +---> nsHTMLToTXTSinkStream ---> TXT part Proposed new code: +---> ScanHTML ------------------------------> HTML part HTML source --+ +---> nsHTMLToTXTSinkStream -----------------> TXT part

> (and what 4.x basically did) In effect, not in code. Since 4.x didn't have so much formatting functionality, both shown versions would (propably) have in the same result.

Ok...after I looked at it again, I have it working the way you describe (in theory). - rhp

Good. Thanks for fixing.

Now working off of original input for HTML-TXT conversion. - rhp

Linux (2000-04-21-08 M16) Win32 (2000-04-21-09 M16) Mac (2000-04-21-08 M16) The plain text part now reads *hello* b@b.com on Won32 and Linux. However, due to bug 35501, html tag exists for plain text on Mac. I am going to leave this bug until bug 35501 is fixed. If you disagree, please advise.

changing qa