Misleading message “Add-on (requires Anki <= 2.1.48)”

After update Anki, say, to 2.1.49, if the currently installed version of an add-on specifies a maximum version (e.g. -48), the add-on will be silently disabled, and in Tools → Add-ons you will see:

Cool add-on (requires Anki <= 2.1.48)

If you start the old version of Anki before updating, which is a fairly normal thing to do, it will try to update, and if the alternative add-on branch starts with a higher version (e.g. 49), it will find no updates. Since Anki checks for updates once a day(?), this means that you will not get the update add-ons prompt after updating Anki.

This certainly will make the user think that the add-on only supports up to Anki 2.1.48. There might be absolutely no hint that in reality there exists an alternative branch that supports cutting edge Anki. This is also confusing for the developers; imagine what issues users create after this!

I suggest:

  • When disabling an add-on, give a warning on Anki startup and offer to update add-ons, even if already checked for updates today;
  • Instead of the above message in the add-on list, say something along

    Cool add-on (You need to update this addon to use it with this version of Anki. The installed version only supports Anki <= 2.1.48)

I think this is already fixed? Add last version check for add-on updates by RumovZ · Pull Request #1608 · ankitects/anki · GitHub

Oh, that’s a little better! However, if the user has an internet problem or something, or if they for whatever reason dismiss the update dialog, the message is still misleading. (This is something I would do—I sometimes prefer not to update add-ons since they tend to break a lot.)

While I still think that the above issue should be addressed, one of my users ran into a problem that might be not related to it. I updated my add-on to have two branches, one for 2.1.0…2.1.-48, and one for 2.1.49…2.1.53. The user seems to have been using Anki < 2.1.49 before, and after updating to 2.1.49 they are not offered the update. I can’t reproduce this on the same Anki using their meta.json. Here’s a rather detailed discussion. Any clues?

1 Like

Weird. Maybe seeing the input and output of fetch_update_info() on their installation would provide a clue.

1 Like