User Agent: Mozilla/5.0 (Windows NT 6.1; Win64; x64; rv:9.0a1) Gecko/20110916 Firefox/9.0a1
Build ID: 20110916071142
Steps to reproduce:
Send WS text message with payload length = 65535.
See test case 1.1.6
Firefox sends a WS text frame starting with
It uses the 7+64 bits encoding.
Firefox should have sent a frame starting with
(when payload isn't fragmented)
This is the minimal length encoding in this case (7+16 bits).
The WS draft 14 says:
5.2. Base Framing Protocol
Multibyte length quantities are expressed in network byte order. Note that
in all case the minimal number of bytes MUST be used to encode the
length, for example the length of a 124 byte long string can't be
encoded as the sequence 126, 0, 124.
Note that the test cases which test payload length 125, 126, 127 and 128 (cases 1.1.2 - 1.1.5) work as expected.
Note that up till now, the test suite I ran until recently a) did not detect overlong payload length encoding and b) the (now failed) case 1.1.6 was doing a payload of length 65536, and thus missed the opportunity to catch these issues anyway. Both of these things are fixed only on Autobahn HEAD (yet unreleased) .. will be in Autobahn 0.4.3.
> The WS draft 14 says:
Hi Tomas, thanks as always for the detailed information. I've been on the road for the last week and haven't had a chance to really dig into your other report (about various utf-8 sequences), but I will do so this week.
As for this issue, Firefox 9 imlpements draft 10, not draft 14. Draft 10 does not have the requirement you cite (as far as I can tell, anyhow.). As such this is a desirable enhancement request and not a bug for this implementation.
(In reply to Patrick McManus from comment #1)
> Hi Tomas,
Tobias, I'm sorry.
Created attachment 560800 [details] [diff] [review]
yep, it was only introduced draft >10, I think only in draft 14, and thus not a bug exactly. Anyway, nice patch .. clearly laid out;)
Try run for 5c4e66ab5a60 is complete.
Detailed breakdown of the results available here:
Results (out of 10 total builds):
Builds available at http://firstname.lastname@example.org
Changing milestone to mozilla10, since the last inbound to m-c merge has already happened, sorry.
This looks like a pretty simple bug fix rather than an issue that requires documentation... any thoughts?
agreed--no doc needed for this.