+++ This bug was initially created as a clone of Bug #570957 +++ See bug 570957 comment 17. The quick fix was the wrong way around. We should have converted the strings to use #1. Replacing %S is just wrong, as it looks like printf and isn't. I.e., if a locale uses %1$S just because they do anywhere else, it breaks. I won't be able to add checks to that one-off, we should fix this to go back to #1. We can do so without entity rev'ing by replacing both #1 and %S in the code, and then the new dashboard will issue warnings. I'd love to get this in ASAP nevertheless, on both sync and firefox repos. Again, sorry for the late notice, just tripped over it.

Moving blocking:beta7+ to ?, because that's nice to beta 7. I'd *really* love to get this in for beta 7 though, in particular before branching, so that we can spare the long tail the noise of this bug. This is an l10n stability fix, and even though we can do it without forcing localizers to update their strings, we should support the standard plurals variabel format quickly, and endorse its use.

Switch the strings from using %S to #1 but support both in the code to cope with outdated strings in other locales. (We're not revving the string names as per Axel's suggestion.)

Comment on attachment 479023 [details] [diff] [review] v1 r=me with nits, we should add more comments. At the call-sites, add a comment that we're using both for historical reasons, and that at some point %S can die? In the localization files, add a comment like # LOCALIZATION NOTE (passwordsCount.label): # Semi-colon list of plural forms. See: # http://developer.mozilla.org/en/docs/Localization_and_Plurals # #1 number of passwords (was %S for a short while, use #1 instead, even if both work) ? That way, we have comments at the relevant places to not accidently remove the wrong parts of the code duplication here.

Add comments as suggested by Axel