See flutter/plugins#6555 (comment) for an example

In that case, what happened was:

• Pre-submit passed, including passing the publishable check of pub publish --dry-run
• 10 days passed waiting for final review
• It landed, and failed publishable in post-submit in pub publish --dry-run.

This is different from the known problem of OOB failures when actually publishing when pub adds new sever-side checks; in this case it appears that the new behavior came in a Flutter roll due to updates in pub. What should happen is that this failure would show up in the roll, but that doesn't happen because pub publish --dry-run doesn't do anything when there's no version change. This is the output from that test in the last roll:

============================================================
|| Running for camera/camera [@0:00]
============================================================
Package camera version: 0.10.0+3 has already be published on pub.

It never tries to run the new analysis, so the new failure didn't show up. I can see two options here:

• See if we can get a new flag added to pub publish for "I know there's nothing to actually publish but let's just pretend, and actually run the checks".
• Change publishable to locally bump the version before running --dry-run so the checks run.

I'd rather we do the former if possible, since any time we're making local changes to what we test we introduce the possibility of false results.