Thread messed up (probably unescaped HTML)

Thread messed up (probably unescaped HTML)

http://my.opera.com/community/forums/topic.dml?id=554161

Latest post in that thread probably has some HTML markup in it that's eating the whole page. Please fix. Thanks

That was me.
I tried to post source of an email message for review. How do I do it? Attachment?

Yeh, the forum post code sucks and always boogers up everything even if you put things in code tags and or escape < and >.

Just open the sent message and ctrl + s to save it as an .mbs file. Then, edit it with a text editor to change any private email address info etc. Then, upload it to My Page -> Files and post the link to the file in the thread.

Shoham, could you send me that e-mail source per PM so we can analyze it, try to reproduce and hopefully fix the problem? Our parser should not give erroneous/invalid HTML, so this is very strange.

If you find stuff like this that breaks the site, please feel free to contact us about it. It’s especially great if you can provide the code that broke the post/site, as this is not intended behaviour at all.

The source code I posted was this -

http://files.myopera.com/Shoham/files/Mail_Source.txt

This is the source of an email with an attachment that was sent from Opera as it was received on the Outlook client. The source is from "view source" in outlook (not "view in a browser" just plain right-clicking the message).

I'd love to know why the attachment's name, a .docx file, gets messed up and appears as ATTXXXXXX(6 digits).dat
I'll repost the question in the appropriate thread once it's up and running again.

This is the source of an email with an attachment that was sent from Opera as it was received on the Outlook client. The source is from "view source" in outlook (not "view in a browser" just plain right-clicking the message).

Actually, I need to see the source of the message in Opera. (As in, go to All Messages/Sent, open the message, right-click on it and choose "View all headers and message". Need *all* of the source.)

Here it is

http://files.myopera.com/Shoham/files/message-Code_Opera.txt

(but it's not what I pasted into the thread of topic 554161)

See:

Content-Disposition: attachment; filename*0="Revised mailshot procedure for Kenes Int";
 filename*1="ernational project teams.docx"
Content-Type: application/vnd.openxmlformats-officedocument.wordprocessingml.document; name*0="Revised mailshot procedure for Kenes Int";
 name*1="ernational project teams.docx"
Content-Transfer-Encoding: Base64

in the source of the message.

You see the filename*0= and filename*1=?

Because the filename is so long, Opera needs to split it up into multiple short values (as header folding won't necessarily work here). The receiving client is then supposed to join the multiple values back together to get the full filename value. The same thing applies to name*=0 and name*=1.

This is covered in: Section 3 - "Parameter Value Continuations" in RFC2231.

The problem is, Outlook doesn't support RFC2231 at all and when it doesn't support a filename value, it uses atxxxxx.

Now, Thunderbird and Sylpheed (by default) violate RFC2231 and send things in a way that Outlook can handle. But, for both Thunderbird and Sylpheed, you can turn on the option to make them conform to RFC2231 so that they do just like Opera. And, I believe with Thunderbird, it will eventually no longer violate RFC2231 by default and will do the same thing Opera does with the filename.

Now, it is unfortunate that Opera doesn't have an option so you can workaround it by violating the rfc. However, the bug is really with Outlook and MS in general as they refuse to support RFC2231 in lots of their products.

So, in short, if you want to send attachments to Outlook from Opera, only attach files with ascii-only filenames and only attach files with filenames that are really short. That will avoid Opera having to use RFC2231 constructs for the filename and name values in the Content-Disposition and Content-Type headers. It will use just simple, short strings that Outlook can handle.

If this is really a big deal, the only thing to do is ditch Opera's mail client for something else that has an option (either default or not) to violate RFC2231.

Hope that helps.

@person fixing that thread: Please add this post to the thread.

If you find stuff like this that breaks the site, please feel free to contact us about it. It’s especially great if you can provide the code that broke the post/site, as this is not intended behaviour at all.

I get problems a lot on the forums and blog posts. But, when it happens, I'm so concentrated on fixing things so that it will actually post right, that by the time I've got it, I've forgotten the exact textarea value that caused the problem. I will try to save these values when I encounter them before messing with them so someone can take a look. I'd love to get this fixed as I'm sick of posting data URI links to avoid this. Will see what I can do next time it happens.

On a side, the forums are really destructive with entities. When you edit your post, your entities are turned into raw text on you. This is especially annoying when you're trying to escape some markup that the forums handle so the forums don't treat it as HTML. We need a "disable markup" option so we can post literal text where the forum properly escapes everything needed before inserting it into the page, but doesn't modify the value we submit. (As in, store the value we submit in the DB as-is and only modify it for the generated page. We should be able to retrieve the original value we submitted when editing or doing a preview.)

Thanks burnout246 for explaining this.
Do you know perhaps what is considered a safe length for the attachment name?
I guess spaces are always a problem.

Do you know perhaps what is considered a safe length for the attachment name?

If you limit the filename (including extension) to 40 ascii characters or less, Outlook should get an attachment filename value that it understands.