User-Agent: Mozilla/5.0 (Macintosh; Intel Mac OS X 10.6; rv:2.0b8pre) Gecko/20101102 Firefox/4.0b8pre Build Identifier: Mozilla/5.0 (Macintosh; Intel Mac OS X 10.6; rv:2.0b8pre) Gecko/20101102 Firefox/4.0b8pre After I complete all the forms form the "Sign up" page and submit my info, the account is not created, even if all the info is correct. Also, I am being redirected to the "Sign up" page without any confirmation or error page, Reproducible: Always Steps to Reproduce: 1.Got to http://www.hotmail.com 2.Click on the "Sign up" button. 3.Complete the required forms and click the "I accept" button. Actual Results: I am redirected to the sign up page, without any confirmation or error message. Expected Results: I should be redirected on a confirmation page if the account was created. I've attached a print screen with the errors I get in the Error Console.
Regression Window: Works: http://hg.mozilla.org/mozilla-central/rev/268ef4ccb5ff Mozilla/5.0 (Windows NT 6.1; WOW64; rv:2.0b7pre) Gecko/20100917 Firefox/4.0b7pre ID:20100917041014 Fails: http://hg.mozilla.org/mozilla-central/rev/bc15c280c430 Mozilla/5.0 (Windows NT 6.1; WOW64; rv:2.0b7pre) Gecko/20100916 Firefox/4.0b7pre ID:20100917044206 Pushlog: http://hg.mozilla.org/mozilla-central/pushloghtml?fromchange=268ef4ccb5ff&tochange=bc15c280c430
Is it known what script ordering property Hotmail's account setup depends on (script-inserted external scripts maintaining order, script-inserted inline and external script maintaining order or script-inserted scripts blocking parser-inserted scripts)? This is UA-sniffed code on Hotmail, right? The mozilla-central script ordering behavior should now match IE/Safari/Chrome. Does Hotmail use LABjs or RequireJS?
If it is _not_ including "Firefox" in UA, the problem does _not_ occur. i.e. set "general.useragent.override" to "Mozilla/5.0 (Windows NT 6.1; WOW64; rv:2.0b8pre) Gecko/20101102 Minefield/4.0b8pre" So, is this the site issue and Tech Evangelism?
Does NOT repro with UA: Mozilla/5.0 (Windows NT 6.1; Win64; x64; rv:2.0b7pre) Gecko/20100916 Firefox/4.0b7pre OR Mozilla/5.0 (Windows NT 6.1; Win64; x64; rv:2.0b7pre) Gecko/20100916 Minefield/4.0b7pre Is this a regression for build 20101102?
(In reply to comment #5) > Does NOT repro with UA: > Mozilla/5.0 (Windows NT 6.1; Win64; x64; rv:2.0b7pre) Gecko/20100916 > Firefox/4.0b7pre > > OR > Mozilla/5.0 (Windows NT 6.1; Win64; x64; rv:2.0b7pre) Gecko/20100916 > Minefield/4.0b7pre > > Is this a regression for build 20101102? Regressed since ID:20100917044206
If Hotmail is depending on script-inserted external scripts executing in insertion order, the problem wouldn't reproduce when the scripts *happen* to load in the insertion order. That might explain the contradictory findings. That is, if you do var s1 = document.createElement("script"); s1.src = "s1.js"; document.body.appendChild(s1); var s2 = document.createElement("script"); s2.src = "s2.js"; document.body.appendChild(s2); The nightlies, just like IE/Safari/Chrome, do NOT guarantee the evaluation order of s1.js and s2.js. Firefox 3.6 did. Bug 602838 has patches waiting for review. If/when those land, sites will be able opt into insertion order execution of script-inserted external scripts by setting .async=false on the script nodes before inserting them. (In reply to comment #4) > So, is this the site issue and Tech Evangelism? Yes, in the sense that Gecko changed to spec-compliant behavior, which is also the behavior already exhibited by IE (since forever) and WebKit, so if the new behavior is causing problems, the logical assumption is that the site is UA sniffing instead of serving cross-browser-compatible code for everyone.
7 years ago
We'll have a nicer evang story once bug 602838 lands.
This isn't about the relative order of script-inserted external scripts. I made a build that made script-inserted external scripts maintain order among themselves, but the signup still didn't work. I suspect the site expects script-inserted external scripts to block subsequent parser-inserted scripts (something the IE and WebKit don't do).
I experimented some more. So far I don't have proof of the signup page depending on the relative order of script-inserted external scripts. On the contrary, the evidence suggests that the page is not depending on that, but I don't have conclusive proof. I do have proof, though, that the page is indeed depending on script-inserted external scripts blocking subsequent parser-inserted scripts. There are two cross-browser and HTML5-compatible ways to fix this: 1) Using document.write("\u003Cscript src='foo.js'>\u003C/script>"); to insert the scripts that need to block the parser. The disadvantage is that when multiple scripts are loaded using this method in one go, the scripts won't load in parallel until bug 543062 is fixed. 2) Simply loading all the necessary scripts by putting <script src="foo.js"></script> tags in the page source. This is the most performant alternative, since Firefox will start fetching the scripts earlier.
(In reply to comment #10) > The disadvantage is that when > multiple scripts are loaded using this method in one go, the scripts won't load > in parallel until bug 543062 is fixed. Bug 543062 is now fixed, so that's no longer a problem.
Microsoft has a fix for this internally. I'll leave this bug open until the fix has been deployed.
It looks like Microsoft has deployed the fix. Thank you!
Verified on Mozilla/5.0 (Windows NT 6.1; rv:2.0b11) Gecko/20100101 Firefox/4.0b11