660522 - Lightning completely replaces the Thunderbird/SeaMonkey user agent in the protocol headers

Status: VERIFIED FIXED

(Calendar :: Lightning Only, defect)

Description

[Stefan Sitter [:ssitter]]

It seems the change from Bug 604680 completely broke the Thunderbird/SeaMonkey user agent. Now the headers only contain "Lightning/1.0b4pre". All information about Thunderbird/SeaMonkey is lost.

Steps to reproduce:
1. SeaMonkey: browse to http://browserspy.dk/useragent.php
2. Thunderbird: set http://browserspy.dk/useragent.php as start page
3. Check the user agent with and without Lightning

Actual results without Lightning: 
Mozilla/5.0 (Windows NT 6.1; rv:5.0) Gecko/20110528 Thunderbird/3.3a4pre
Mozilla/5.0 (Windows NT 6.1; rv:2.0.2pre) Gecko/20110526 Firefox/4.0.2pre SeaMonkey/2.1.1pre

Actual results with Lightning:
Lightning/1.0b4pre

Expected results:
Lightning should not replace the user agent. If necessary it should just append its own string.

Comment 1

[Philipp Kewisch [:Fallen] [:📆][:🧩]]

Attachment: Fix - v1

This patch takes care. It seems User-Agent is no longer a mergeable header, or it never was and I made a mistake :) I'd appreciate review.

Comment 2

[Tony Mechelynck [:tonymec]]

Mozilla/5.0 (X11; Linux x86_64; rv:7.0a1) Gecko/20110601 Firefox/7.0a1 SeaMonkey/2.2a1pre ID:20110601003002

with Lightning 1.1a1pre nightly dated 2011-06-01

I see this too (on Linux x86_64): the site mentioned in comment #0 tells me:
- UserAgent from header 	Lightning/1.1a1pre
- User Agent from JavaScript 	Mozilla/5.0 (X11; Linux x86_64; rv:7.0a1) Gecko/20110601 Firefox/7.0a1 SeaMonkey/2.2a1pre
- Browser string 		lightning
- Operating System 		Unknown
- Browser name 			Yes - version firefox
- Browser version 		Yes - version 7 

=> Platform to All / All

Comment 3

[Stefan Sitter [:ssitter]]

Comment on attachment 536067 [details] [diff] [review]
Fix - v1

r=ssitter

Comment 4

[Philipp Kewisch [:Fallen] [:📆][:🧩]]

Pushed to comm-central <http://hg.mozilla.org/comm-central/rev/4f9afaea10c9>
-> FIXED

Comment 5

[Philipp Kewisch [:Fallen] [:📆][:🧩]]

Backported to comm-miramar <http://hg.mozilla.org/releases/comm-miramar/rev/b917de5e8311>

Comment 6

[Ian Neal]

This should probably go into comm-aurora and comm-beta too?

Comment 7

[Philipp Kewisch [:Fallen] [:📆][:🧩]]

TBH I need to find a strategy for the additional branches first. Lightning can't handle builds for three branches, and we can't be committing to 4 branches for each patch...

Comment 8

[Tony Mechelynck [:tonymec]]

Lightning tinderbox-build downloaded from ftp://ftp.mozilla.org/pub/calendar/lightning/tinderbox-builds/comm-central/linux64-xpi-linux64/1307121907/lightning.xpi dated 2011-06-03 12:14 gives me the same results as in comment #2.

I see the next hourly is up, let's try again.

Comment 9

[Tony Mechelynck [:tonymec]]

Mozilla/5.0 (X11; Linux x86_64; rv:7.0a1) Gecko/20110602 Firefox/7.0a1 SeaMonkey/2.2a1pre ID:20110602231050

With the hourly from subdirectory 1307128485/ datestamped 13:33 the results are better, as follows:

Test 				Result
UserAgent from header 		Mozilla/5.0 (X11; Linux x86_64; rv:7.0a1) Gecko/20110602 Firefox/7.0a1 SeaMonkey/2.2a1pre Lightning/1.1a1pre
User Agent from JavaScript 	Mozilla/5.0 (X11; Linux x86_64; rv:7.0a1) Gecko/20110602 Firefox/7.0a1 SeaMonkey/2.2a1pre
Browser string 			Mozilla/Netscape 6.x
Operating System 		Linux/Unix
Browser name 			Yes - version firefox
Browser version 		Yes - version 7 

(1) As you can see, the Lightning version is appended to the HTTP header but not to the Javascript result (and, as I checked, not to the value shown by about: or inserted at top of this comment by the Nightly Tester Tools extension).

(2) Also, I have tested only the Linux x86_64 version but I assume that this part of the Lightning code is platform-agnostic.

I'm marking this bug VERIFIED on the assumption that these are the desired results. Please REOPEN if (1) above is not exactly what was desired, or if the same result is not observed on a comm-central build datestamped "late enough" after comment #4.

If a comm-miramar build datestamped "late enough" after comment #5 fails the test, then IIUC the fact should be mentioned on the Whiteboard without reopening the bug.

For the meaning or "late enough", compare comment #8 with this one.