The upload-import path never increments duplicates (a dedup hit
returns Ok and counts as success), so the counter and the progress
math addition could only ever render dead zeros. The display returns
together with the server-side fix; the client-side aggregation of the
duplicates field stays, as issue #2 specifies.
The import footer showed 'Will import to: X' with no sign that the
whole selection lands in that one folder, and the detected-folder
badge never said which file the hint came from (only the first valid
file is consulted). Multi-file selections now show the file count in
the footer and the hint's source file in the badge.
The results card now renders the aggregated duplicates count and the
processed math includes it. The upload-import path currently always
reports duplicates as 0 (duplicates return Ok and count as success
server-side), so this only becomes visible once the server reports
them, but the aggregate and display are ready.
Header unfolding in extractFolderHint matched any whitespace-led line
in the remaining 64 KB, so body text could be glued onto the folder
name; it now stops at the first non-continuation line per RFC 5322.
The file picker allows multi-select but the import mutation only ever
uploaded files[0], silently dropping the rest. upload-import is
one-file-per-call and async: it returns Pending immediately and the
real counts only exist at /import-progress/:id, so summing upload
responses would aggregate zeros.
importFiles() loops the selection sequentially: upload, poll to a
terminal status, merge counts and failed_details into an aggregate.
A file that fails upload or polling becomes a synthetic failure entry
(labelled with the file name) instead of aborting the remaining
uploads; only when every upload transport-fails does it throw so the
existing toast-and-reset path still handles total failure. Upload
progress is byte-weighted across the whole selection so the bar never
resets between files. An AbortSignal wired to component unmount
replaces the deleted setInterval cleanup so polling cannot outlive
the page.
Fixes#2
- Add SyncFull trigger and configurable IMAP socket read timeout
- Replace streaming UID FETCH with UID SEARCH ALL + batched fetch so per-message progress stays responsive and throttling servers can retry with reconnection
- Finalize stale Running sessions on startup so interrupted syncs no longer show a phantom "syncing" state
- Show a live syncing pill on the account row; add elapsed time, current folder, and slow-server warning styling in the dialog
- use a single Engine instance for email + attachment blobs
- add delete_batch, gc_if_needed, background flush to blob crate
- fix Entry.raw_size storing compressed length instead of original
- move fjall-dependent migration code from core to admin crate
- add STORAGE_VERSION file for layout version detection
- Replace bincode 3.0.0 (empty crate) with bincode_reloaded 3.1.10
- Add delete_batch for efficient grouped tombstone writes
- Add gc_if_needed to skip GC when no segment exceeds threshold
- Add flush() for lightweight fsync+meta checkpoint without compact
- Add Config::flush_interval_secs to spawn a background flush thread
- Remove per-put/per-delete fsync; persistence via background flush
- Split gc_segments into gc_prepare/gc_finish to reduce write lock hold time
