I am very scared to use BFs with a copy of my email. Even worse if it was configured to do partial downloads. Why? Because that is confusing and a risk of losing data. It is never clear if an operation on the local filesystem will be properly translated to the online one. Was the mail daemon connected to the server when you did the operation? Did it pick the filesystem notification? Did it have time to synchronize it with the server before it was disconnected or the machine was shut down? If I move a mail from my inbox to another directory, will it be removed from the IMAP server as well?
And in the other direction: if I delete mails from my inbox, will it delete the copy on my hard disk as well, or can I trust that as a backup? If I mark emails as read, will it propagate?
This asynchronous synchronization happening in the background can’t be reliable. There is no indication that the synchronization is off (I would want something that put the inbox readonly or at least clearly marks that the changes are not synchronized). You wouldn’t want to work that way for other networked filesystems like NFS or SMB, and that’s how I think about my email inbox: a shared network resource that I can access from various computers.
So, if I want to store an email on disk, I’d simply comy it from the mounted IMAP directory to one of my local drives. This happens synchronously (with a Tracker file copy dialog showing progress). Same thing if I decide to delete emails, move them around, etc.
The filesystem may then decide to optimize a few things with a local disk cache. But that tricky part becomes a filesystem problem, something that, as a user, I don’t have to care about. With the mail daemon, this is not at all the case. I have to think constantly about the fact that there is a remote IMAP mailbox, and a local copy, and that they are somewhat synchronized, but not really, and that any operation may or may not replicate between the two.
So, yes, independendtly of the UI, I don’t consider the current approach safe enough to trust my email with it. And that is not a UI problem.
But that’s just me and the way I use my email. I understand that other people are happy with it. As humdinger says, if you think about it in the POP protocol spirit, where you download mails to your machine and then there is no synchronization to the mail server, that approach works great.
However, this is the one nice thing about representing email as single files with attributes: both of these approaches can be implemented. The backend can be different (a server running in the background, or a filesystem), and we could still all use the same email client frontend/UI to manage our emails.