The .com domain is not as IDN clean as we believed. We need to back out the whitelisting done in bug 770877 and rely on the IDN2008 algorithm in bug 722299 instead. This will fix bug 840882. This is sort of a dupe of that one but thought I'd file it separately for clarity of action and decision. There may be additional work that could take place in bug 840882 and this could be a quick solution to return to the pre-Firefox 17 status quo. Or this approach could be vetoed in which case a separate bug records that decision better than buried in the comments of a bug that ultimately goes another way.

I think we should do this in the same release that we ship bug 722299. To do it earlier than that means that a large pile of perfectly legit IDN domains which have worked for a period would stop working for a release or two, then start working again. Given that Verisign have tightened up their registration rules so it's currently not possible to register domains such as the ones used in 2005, I think this is acceptable. Gerv

Is this something that needs to ride the trains?

(In reply to Gervase Markham [:gerv] from comment #1) > I think we should do this in the same release that we ship bug 722299. To do > it earlier than that means that a large pile of perfectly legit IDN domains > which have worked for a period would stop working for a release or two, then > start working again. bug 770877 only surfaced in Firefox 19 so if this bug on it's own (simple pref change) jumped the trains to Firefox 20 that's not a very long period for those domains to have worked before breaking again. There are two classes of such domains. Those that were first registered before the Firefox 19 release were created when they were broken in Firefox; that they worked for a time was a bonus for those owners but not anything they counted on. The only ones I'd really feel bad about are those who, sometime after we released Fx19, noticed the change in our behavior and decided to buy an IDN domain based on that knowledge. That can't be that many domains. Clearly Firefox is best with bug 722299 in addition to this bug, but even just this bug fix is an improvement for most of our users. It's a negative for a few--and especially Verisign--in the very short term. My main worry is that bug 722299 won't make Aurora (Firefox 21) and then we'll be living with potential problems until Firefox 22. I wouldn't want this fix to wait until Fx22, and if there's going to be a period of it landing without bug 722299 we might as well fix things in Firefox 20 rather than wait and let another 6 weeks worth of IDN .com domains be bought before breaking.

Sorry, I wasn't CCed on this bug. I have input here, but http://blog.gerv.net/2013/03/john-phinehas-markham/ has just happened; I hope to get back to work soon. :-) The next migration isn't imminent, so I hope you guys are OK to wait a few days. Gerv

Currently, the fix for bug 722299 is undergoing some additional and specific testing. If this passes and we decide to uplift to Aurora, I think we should consider avoiding a compatibility break. If we don't uplift, I think there's more of a case for checking this part in (and breaking .com IDN again) early. As a veteran of the UA wars, I'm wary of making breaking compatibility changes without data. And I suspect we may not get IDN .com sales data out of Verisign... Gerv

Is there a plan to uplift the patch to Beta?

The current plan is that this bug and bug 722299 are riding the trains together, without additional uplift. Gerv

(In reply to Gervase Markham [:gerv] from comment #12) > The current plan is that this bug and bug 722299 are riding the trains > together, without additional uplift. > > Gerv Thanks Gerv, I am marking status-firefox21: wontfix & let this ride the trains and get it fixed in FX22 timeframe.

bhavana: Hmm. I thought bug 722299 and this bug were going to make Fx21 (and therefore the ESR). Given that they are not, I might apply for an uplift. Gerv

(In reply to Gervase Markham [:gerv] from comment #14) > bhavana: Hmm. I thought bug 722299 and this bug were going to make Fx21 (and > therefore the ESR). Given that they are not, I might apply for an uplift. > > Gerv Hi Gerv, I marked this as wontfix considering comment #12.But if we need to uplift , looks like you have nominated all patches in 722299 which will be considered in today's triage.