mergetag appears if you're merging signed tags with a newer version of git. That's indeed less of a conscious decision than e.g. using submodules. A lot of git development happens without signed tags though, and bzr-git has never supported signed tags (it just ignores them). There are a bunch of other things that have broken existing imports too - using a filename with a backslash in it, for example, and there are some repositories that introduced oddly formatted file modes written by certain git implementations (bug 581064) .
Most imports are actually created by upstreams that don't care about the Bazaar imports on Launchpad. Both submodules and mergetag are regressions from the point of view of whoever is relying on these imports in Launchpad.
There are at least as much imports on Launchpad that started failing because their upstream started using submodules as imports that fail because of mergetag.
I'm not opposed to bumping this to critical - there seem to be a fair number of users who are affected by it - unlike e.g. bug 581064. I do think we should be consistent though if this is labeled a regression.
mergetag appears if you're merging signed tags with a newer version of git. That's indeed less of a conscious decision than e.g. using submodules. A lot of git development happens without signed tags though, and bzr-git has never supported signed tags (it just ignores them). There are a bunch of other things that have broken existing imports too - using a filename with a backslash in it, for example, and there are some repositories that introduced oddly formatted file modes written by certain git implementations (bug 581064) .
Most imports are actually created by upstreams that don't care about the Bazaar imports on Launchpad. Both submodules and mergetag are regressions from the point of view of whoever is relying on these imports in Launchpad.
There are at least as much imports on Launchpad that started failing because their upstream started using submodules as imports that fail because of mergetag.
I'm not opposed to bumping this to critical - there seem to be a fair number of users who are affected by it - unlike e.g. bug 581064. I do think we should be consistent though if this is labeled a regression.