pruning does too much work

Currently pruning the database will delete records of files that do not exist and also rehash every image that still does. The second step is both costly and unnecessary in a simple workflow:

1. generate phashes for all images in a large library
2. get list of dupes
3. delete some dupes
4. remove deleted dupes from database

and also rehash every image that still does

Not exactly. It rehashes only those files which have their modification time changed.

Hmm, yes, I see. Maybe my timestamps aren't comparing as expected. When I wrote this, I'd just killed the process after 90 minutes of CPU time following a delete of ~1k files from my photo storage.

I added some log statements and discovered that I'm hitting this error case on every file. It really is recalculating every hash.

Incidentally, all of the lastmod values are bizarre:

sql: Scan error on column index 2, name "lastmod": converting driver.Value type time.Time ("1921-05-26 23:09:49.266591744 +0000 UTC") to a int64: invalid syntax

sql: Scan error on column index 2, name "lastmod": converting driver.Value type time.Time ("2133-09-18 06:16:29.266591744 +0000 UTC") to a int64: invalid syntax

I haven't taken any pictures in 1921 or in 2133. Yet!

closed via commit 5ad82a3e

Fixed, thanks for the report. Old databases will need to be regenerated, though.

added bug label

Thanks for the software!