13.70 KB, application/octet-stream
** Observed with 6/16/99 Win32 build ** Here's a part of a UTF-7 mail sent from OutlookExpress: +ADwAIQ-DOCTYPE HTML PUBLIC +ACI--//W3C//DTD W3 HTML//EN+ACIAPg- This is how it is displayed in Messenger 5.0 above. Communicator 4.6 does not display this at all since it is part of the HTML structure, abd that is correct. Here, UTF-7 Set O (optional direct characters) are encoded by Outlook Express, but we apprently expect all Set 0 characters to be directly represented. We should be able to deal with optionally encoded characters also.
Here's additional info. I took the above UTF-7 string and put that into an unlabeld .txt file and placed it at the above URL: 1. http://rocknroll/users/momoi/publish/seamonkey/tests/optcharutf7.txt Also I replaced the optionally encoded characters with direct representations and placed the string in a .txt file at: 2. http://rocknroll/users/momoi/publish/seamonkey/tests/optcharutf7b.txt Under Communicator 4.6, I can see both of them correctly under UTF-7 as: <!DOCTYPE HTML PUBLIC "-//W3C//DTD W3 HTML//EN"> With 5.0 Browser, under Latin 1 encoding, I see +ADwAIQ-DOCTYPE HTML PUBLIC +ACI--//W3C//DTD W3 HTML//EN+ACIAPg- for URL 1 but NOTHING for URL 2. Under UTF-7. I see NOTHING for either URL 1 or URL 2. These are text files and we should be able to show the string under either Latin 1 (incorrectly) or UTF-7 (correctly).
I just checked. Actually the UTF-7 decoder is accepting anything base64-encoded, including Set O. So, the converter seems to be ok. About the two test URLs: they are txt files with HTML inside. But that HTML is only a header, so of course we don't show anything in the page. If there is a problem, that is that the txt is parsed as HTML. But the very fact that for the URL 1 with encoding set to UTF7, we don't show anything, proves that the converter is working right: the text got converted into that HTML header!!!!!! Adding a single valid HTML tag in there puts something visible in the page, proving once again that the converter is ok. Now, about the original issue. Tested in browser (that piece of encoded text) works. So I guess it is some other problem. Maybe the encoding is not setted right in the mail? I do not know. But we should reassign this bug to the right owner.
OK. Let's send this over to nhotta then. What should we do about the .txt file interpreted like .html file in 5.0? A separate bug?
I was able to see UTF-7 encoded Latin1 character mailed by OE5 (attachment created). We need more data for this (original bug was filed with OE4, we also need 4.x data). Viewing UTF-7 is not a not a major requirement for M8. Moving to M10.
Marking as FIXED. I saw 6983 was fixed and verified. With the converter's fix, this should be resolved now.
** Checked with 7/12/99 Win32 build ** The original problem, i.e. some HTML structures were displayed rather than suppressed, is now gone. So in this sense, the bug has been fixed though I don't know what check-in has solved the problem. The 2nd problem is that of mistaking .txt files as if it were .html files. This is not directly relevant to the UTF-7 coverter and will be filed as a separate bug. Marking it verified/fixed.