Spanish vodafone registration page doesn't work because of illegal character but it works in other browsers
Categories
(Core :: Networking, task)
Tracking
()
People
(Reporter: emilio, Unassigned)
References
()
Details
Opening the URL above shows just a blank page in Firefox, but works in other browsers.
We throw a bunch of Uncaught SyntaxError: illegal character U+001F
, which seem to prevent the scripts in the page from executing, but other browsers do make progress (they perhaps ignore it?).
Reporter | ||
Comment 1•3 years ago
|
||
There's also a bunch of warnings like:
The script from “https://www.vodafone.es/c/mivodafone/es/registro/polyfills.701f3d0ef21eae12ab6b.js” was loaded even though its MIME type (“application/x-gzip”) is not a valid JavaScript MIME type.
Which perhaps is what's going on? Are we not uncompressing the scripts but other browsers are?
Reporter | ||
Comment 2•3 years ago
|
||
Raw headers from one of the problematic scripts:
HTTP/1.1 200 OK
Date: Fri, 01 Oct 2021 10:02:34 GMT
Server: Apache
Strict-Transport-Security: max-age=60000; includeSubdomains;
Vary: Accept-Encoding
Last-Modified: Thu, 30 Sep 2021 05:57:38 GMT
Accept-Ranges: bytes
Content-Length: 30821
Cache-Control: max-age=0
Expires: Fri, 01 Oct 2021 10:02:34 GMT
Access-Control-Allow-Headers: auth_ticket,userdata
Access-Control-Allow-Credentials: true
Access-Control-Allow-Methods: *
Content-Encoding: gzip
Keep-Alive: timeout=5
Connection: Keep-Alive
Content-Type: application/x-gzip
Reporter | ||
Comment 3•3 years ago
|
||
Perhaps this is more of a necko / DOM-y thing? Valentin, thoughts?
Comment 4•3 years ago
|
||
Yes I think we don't decompress the gzip contents. The illegal character, 0x1f, is the first byte of the gzip magic number.
Comment 5•3 years ago
|
||
This seems related to bug 1030660. If I comment out the code in nsHttpChannel::ClearBogusContentEncodingIfNeeded
the website works...
Reporter | ||
Comment 6•3 years ago
|
||
Nice find!
Comment 7•3 years ago
|
||
Thanks for fixing that Emilio!
Description
•