dwitte said he would own this, so assigning to him

Yup, I'll go through a bunch of UA bugs we've got open tomorrow, and get things rolling...

OK, listed dependencies are all the ones I can find that are currently on file that I think are relevant. (Some will be WONTFIX; listing for completeness.) If we took all those changes, the UA would end up looking something like: Gecko/2.0b4pre (Linux x86_64) Minefield/4.0b4pre Gecko/2.0b4pre (Windows NT 5.0) Minefield/4.0b4pre Gecko/2.0b4pre (Intel Mac OS X 10.5) Minefield/4.0b4pre And for stable releases: Gecko/1.9.3 (Linux x86_64) Firefox/3.6 Gecko/1.9.3 (Windows NT 5.0) Firefox/3.6 Gecko/1.9.3 (Intel Mac OS X 10.5) Firefox/3.6 I'm fully expecting that we won't go that far. We can discuss in the relevant bugs, lest this bug turn into noise.

(I missed bug 588874 in the UA strings above; s/Minefield/Firefox/.)

This is probably beyond the scope of this bug, but if someone were to write a static library and then encourage others to use a remote-js include of that library on their sites, then there wouldn't be so much breakage with changing UA strings since updating the static library would affect the sites. Maybe an idea for the future.

Talked with jst, blizzard, and dveditz, and (mostly) settled on what we're doing. I have some more bugs to file on platform-specific changes.

OK, dust has settled. I'm going to write a blog post today detailing what we're doing for 4.

(In reply to comment #6) > OK, dust has settled. I'm going to write a blog post today detailing what we're > doing for 4. Bug 581008 and related are UA string handling changes, rather than string changes themselves, though they will affect a lot of existing UA strings. It might be good to announce everything at once and get bug 581008 done asap.

Done.

> Talked with jst, blizzard, and dveditz, and (mostly) settled on what we're > doing. While I respect the decision, I would like to see specific reasons, for the individual bugs that were WONTFIXed. Maybe, for FF5, there are ways to resolve whatever concerns there were now, with testing, advocacy, documentation or workarounds.

(In reply to comment #9) > While I respect the decision, I would like to see specific reasons Absolutely -- been meaning to post more detail, but have been busy this week. (Beta freeze is now.) I'll post more in the various bugs next week, as well as a second blog post with more detail.

Is it intentional that the Firefox 4.0b5 candidate 1 builds contain Gecko/20100101?

Yep!

(In reply to comment #11) > Is it intentional that the Firefox 4.0b5 candidate 1 builds contain > Gecko/20100101? Yes. See bug 592342 and bug 591537. Minefield and branches should be unaffected.

Minefield and branch nightlies and hourlies should still have build IDs, I mean.

Posted more detail on rationale in the various bugs.

This will need some doc love. Sheppy, feel free to coordinate with me if you need a rundown on what should be updated. I'm currently updating https://developer.mozilla.org/en/Gecko_user_agent_string_reference in preparation for a blog post thismorning, so that part will be taken care of. (Though please do edit it as you find necessary.) In particular, https://developer.mozilla.org/en/Browser_Detection_and_Cross_Browser_Support will need changing. Some of the recommendations there are obsolete, or non-futureproof.

We should also delete (or redirect to https://developer.mozilla.org/en/Gecko_user_agent_string_reference) https://developer.mozilla.org/en/Gecko_User_Agent_Strings. The information there is very sparse and not particularly useful.

Hmm. Bugzilla ate my keywords. (See previous two comments.) For relnote, we have Exhibit A: http://hacks.mozilla.org/2010/09/final-user-agent-string-for-firefox-4/ Exhibit B: https://developer.mozilla.org/en/Gecko_user_agent_string_reference Exhibit C: http://people.mozilla.com/~dwitte/ua.txt I think a one-liner along the lines of "The User Agent string has changed for Firefox 4, which may require updates to websites that use the string. See http://hacks.mozilla.org/2010/09/final-user-agent-string-for-firefox-4/ for further details." would suffice.

> http://hacks.mozilla.org/2010/09/final-user-agent-string-for-firefox-4/ Replied in bug 588909 comment 23.