While I guess I can see how in Mozilla's mind the automatic connections are not 
a bug, the documentation bug is undeniable. A documentation page contains 
information that is simply inaccurate and incomplete. I believe that they 
closed this bug prematurely just because you pissed them off. I finally managed 
to partially replicate some of your results, which I have shared along with a 
request to reopen the bug.

I posted as soon as I had some results to show because I did not want to delay 
any longer, but I am still struggling to fully replicate your results. Would 
you mind clarifying some things for me?

My procudure was to reboot, start tcpdump, open Firefox (having started with a 
fresh download of Firefox 58 and followed the steps on the documentation page 
in question), close Firefox, and stop tcpdump. Your tcpdump command as 
specified in the bug report was

# tcpdump -i eth1 ip src host pc and dst host not router and dst host not pc 
-ltq

Since I am using WiFi and not Ethernet I changed it to

# tcpdump -i wlan2 ip src host pc and dst host not router and dst host not pc 
-ltq

which gave me "tcpdump: unknown host 'router'", so I removed 'and dst host not 
router'

# tcpdump -i wlan2 ip src host pc and dst host not pc -ltq

This gave no errors or warnings, but it also did not show any connections. I 
had to remove the boolean expressions:

# tcpdump -i wlan2 -ltq

To get any output, which showed connections to cloudfront.net. After removing 
the arguments '-ltq'

# tcpdump -i wlan2

and examining the output I was able to find some additional urls among the full 
output, which I listed in my comment requesting that the report be reopened.

So while I can verify that Firefox makes some automatic connections, your 
results seem a little different from mine, and I'd like to figure out how to 
replicate more of your results so that I can make a stronger argument in the 
Mozilla thread.

Some differences between my procedure and yours are that I am using Trisquel 8 
instead of OpenSUSE and i3 instead of Plasma, but I do not see a particular 
reason why this should make a difference. It is more likely that I am doing 
something wrong because I am not as proficient with tcpdump as you are. It is 
also possible that I am confusing this specific test with some of the other 
situations you have tested and that I need to try more situations that simply 
opening and closing the browser (robots.txt etc.) I have not tried these yet 
because I am very pressed for time at the moment and believe that showing 
automatic connections on startup and closing should be enough to justify 
reopening the bug report, but I can try them if you think it will significantly 
strengthen our case.

Reply via email to