feat: split envelopes by time of arrival

[GretaD]

fixes #11057

to do

• do not show an empty section

[ChristophWurst]

For reactivity: would it make sense to maintain the brackets in Pinia? Brackets relative to current time are probably not reactive. But if we commit the timestamp to Pinia, e.g. at every synchronization, and re-evaluate, it should be fine.

[st3iny]

There is a problem with an empty view initially. When the app loads there are no envelopes at first. Fetching initial envelopes is handled inside Mailbox.vue which is not rendered as there are no envelopes initially (and no groups as a consequence). As a result, no envelopes will be fetched and rendered until the first background sync hits.

My suggestions:

1. Move the sections/groups from MailboxThread.vue into Mailbox.vue.
2. Move some of the envelope fetching logic from Mailbox.vue to MailboxThread.vue (have a look at mounted(), loadEnvelopes() and prefetchOtherMailboxes() in Mailbox.vue).

In any case, you should rebase against latest origin/main soon as I implemented a lot of changes to the sync logic in Mailbox.vue.

[ChristophWurst]

For reactivity: would it make sense to maintain the brackets in Pinia? Brackets relative to current time are probably not reactive. But if we commit the timestamp to Pinia, e.g. at every synchronization, and re-evaluate, it should be fine.

Bump. You have a computed property that depends on time. How does Vue handle this side effect? Is the window for messages in the last hour evaluated again as time passes?

[GretaD]

For reactivity: would it make sense to maintain the brackets in Pinia? Brackets relative to current time are probably not reactive. But if we commit the timestamp to Pinia, e.g. at every synchronization, and re-evaluate, it should be fine.

Bump. You have a computed property that depends on time. How does Vue handle this side effect? Is the window for messages in the last hour evaluated again as time passes?

you are right, the current one is not the ideal solution, i will do what you suggested and commit the timestamp to Pinia. I was focused on fixing the bugs i caused with this feature and didnt pay attention to this part yet.

[kesselb]

I will suggest a different approach later because I find it a bit weird to use a translate string as key.

[GretaD]

hey, i pushed another fixup, is that what you had in mind? I dont mind changing it again if you have a better approach.