Snapshot of the POP3 sync + storage model, with emphasis on UIDL vs Message-ID and why it matters for lookup. See Database for the general GDBM schema.
• Current Storage — What's in each GDBM file and how it's keyed
• UIDL vs Message-ID — What they are and why it matters
• The Lookup Tension — Why Message-ID lookup currently requires scanning
• BOM Issue — BOM-prefixed stored messages breaking header parsing
• Planned Indexes — Proposed Message-ID index and other improvements
Optical RX packet lock not achieved yet on latest capture; decoder transitioned to Python-first path for speed. Current best evidence: stable near-lock clusters and 0x7E occurrences but no AA+55 header recovery.
Latest decode_fast_py pass (17,280 trials) still shows no AA+55 header lock. Strong recurring 0x7E-only signatures remain, especially in inverted edge-mode hypotheses at ROI 316,96,96,40.