864084 - It tries to parse as xml, but this is an archive

Status: RESOLVED INVALID

(Firefox :: Untriaged, defect)

Description

[lucian.jantea]

User Agent: Mozilla/5.0 (Windows NT 5.1; rv:19.0) Gecko/20100101 Firefox/19.0
Build ID: 20130307023931

Steps to reproduce:

When you click on any link containing the word "xml" anywhere in it, then Firefox will think that this is an xml page. How clever (dumb) is this for a browser?

Example, I want to download the following COMPRESSED .BZ2 archive by clicking on the link dirrectly:
http://dchublist.com/hublist.xml.bz2

Note, the same error will happen if you enter at the address bar dirrectly the http://dchublist.com/hublist.xml.bz2 and press "Enter"

The same thing will happen at all DC++ sites.



Actual results:

XML Parsing Error: not well-formed
Location: http://dchublist.com/hublist.xml.bz2
Line Number 1, Column 8:BZh91AY&SY?8xq



Expected results:

It would be nice for a pop-up dialog to appear and ask me: "Save As...."


How can I disable this silly in-address scanning for the word "xml"?

Comment 1

[Loic]

Same error with IE9 (blank page).

Comment 2

[Matthias Versen [:Matti]]

via http://web-sniffer.net/ and the URL http://dchublist.com/hublist.xml.bz2

Status: HTTP/1.1 200 OK
Server:	nginx	
Date:	Sun, 21 Apr 2013 11:50:44 GMT	
Content-Type:	application/xml	
Connection:	close	
Last-Modified:	Sun, 21 Apr 2013 11:50:02 GMT	
Accept-Ranges:	bytes	
Content-Length:	45543

The server sends us a application/xml document. 
There is no scanning involved, we use the content-type as every browser has to do.

Comment 3

[Matthias Versen [:Matti]]

I forgot to add the workaround for this server bug: shift+click on the link

Comment 4

[lucian.jantea]

Unfortunately the server will send back data starting with "BZh91AY&SY?8xq" and everyone can recognize the Bzip signature in the first bytes of the response.

Shift+click on a link just open it in a new window where will throw the same error instead of opening a "Save as.." dialog.

More examples of this xml behaviour:
http://www.stealthsettings.com/hublist.xml.bz2

While searching for more examples, I could see that in many other cases (for other websites), the "Save as.." dialog will appear ok.

All the above do not change this:
Thank you for a great browser wich I use everyday.

Best regards.

Comment 5

[Matthias Versen [:Matti]]

>Unfortunately the server will send back data starting with "BZh91AY&SY?8xq" and 
>everyone can recognize the Bzip signature in the first bytes of the response.

The magic number or the file extension don't matter if the server sends an explicit http content type header.
That's mandatory by the http rfc and is also a security feature. IE had some security issues in the past due to not following the content-type header and doing content-sniffing (e.g. html/js uploaded as image on some site that allow uploading images allows to steal a session cookie).

Comment 6

[Masatoshi Kimura [:emk]]

(In reply to Loic from comment #1)
> Same error with IE9 (blank page).

And also Chrome and IE10. Even a compat issue is not present.