This comment illustrates a case where the user left an incomplete draft without knowing the draft remain saved and thought the work was lost.

We need to make sure that the notification is sent when there is a high confidence that the user left the draft editing and it is not conflicting with the notification of first-time publishing (e.g., the 1st publish notification replaces it if it was not read)

@santhosh do you know if there is a way for sending this notification when the user leaves the 1st translation incomplete in a reasonable reliable way?

The information we need to accomplish this are

1. In a translation session - the number of translations that the user did so far - cxtranslatorstats API(example) will give that .
2. When user is leaving the translation with 'in progress' status, if the translation count ===0 , we need to add a notification ContentTranslationStorage.prototype.onPageUnload is the handler for the bodyunload. But I dont see a way to trigger a notification from client side. Anyway, doing stuff on window.bodyunload is not nice and not reliable.

Alternatively, we can have a scheduled maintainance job that does this: For users with 0 translations and time difference between last updated unpublished translation and now is 10 hours(for example) we can trigger that notification, The notification should not sent more than once. Such a maintainance job can be expanded to remind translators about translations that are in draft status for months or so.

We could also insert a delayed job (Job::getReleaseTimestamp) when a draft is saved and published count is zero. With appropriate de-duplication of course.

Such a maintainance job can be expanded to remind translators about translations that are in draft status for months or so.

That would help with T89707: Encourage users to publish old inactive unpublished translations, but for that specific case we need to have some additional considerations: (a) make sure that very old drafts load fine, and (b) determine what to do for old drafts for which target article now already exists (which we don't want to encourage overwriting with an old version).