See bug 274481. I changed some things in the bug, namely I resolved it and moved it to another product. What I did not change was the review flag of attachment 168713 [review]. Nevertheless bugzilla ate the r+ flag Neil gave this patch before. I'm adding the dependency to bug 274366 just because this happended shortly after the bmo upgrade.

in bugmail, this appeared as: Flag|review+ | as if a bug flag (rather than an attachment flag) was removed...

What were the source and destination products? Several products have their own review flags, and most flags are typically only valid in one or two products. Moving a bug out of the product it was flagged in is very likely to remove the flags because they're no longer valid in the destination product. It was this way before we upgraded, too.

*** This bug has been marked as a duplicate of 261995 ***

(In reply to comment #2) > Several products have their own review flags, and most flags are typically only > valid in one or two products. Moving a bug out of the product it was flagged in > is very likely to remove the flags because they're no longer valid in the > destination product. We should prevent that. Two products with flag types having the same name should prevent flags from deletion. If I cannot find a bug about that, I will open one.

Bug 274481 was moved from Firefox to the PSM product. The review+ flag on an attachment is valid in both products. This is unlike the Toolkit product with its first-review and second-review (did anybody ask for those btw?) flags. Reopening.

Steffen, did you read my comment 4? You can have a flag type "review" in a product which is not the same flag type "review" in another product. For *you*, they are the same, but for Bugzilla, they are different!

Sorry, that wasn't clear to me. So bugzilla allows flags with equal names to be different across products? What is the reason for this? Why should review+ in Firefox be different from review+ in PSM?

Here is an explanation. You can CC different people by default for different products, which really makes sense: justdave: yes, there are separate review flags for different products (because they CC to different people and so forth) justdave: review in the bugzilla product CCs reviewers@bugzilla.org on all review requests, for example justdave: and I think some of the Mozilla-related ones have it CCing netscape.public.mozilla.reviewers or something like that Moreover, now that bug 180879 is fixed, you can say that for the product A, only the group 1 is allowed to request or grant/deny flags, while in the product B, you want to allow another group 2 to request or grant/deny flags. Very useful too! Steffen, I opened a bug about this problem, which is reasonable to fix. See bug 274802. *** This bug has been marked as a duplicate of 261995 ***

Thanks Frederic, Dave just explained it to me as well on IRC. (Although I don't understand why this bug isn't a dupe of your newly filed bug 274802, but never mind.)