only calculate a document's CRC32 if needed

koreader / crengine

This is the KOReader CREngine fork. It cross-pollinates with the official CoolReader repository at https://github.com/buggins/coolreader, in case you were looking for that one.

72 stars 45 forks source link

only calculate a document's CRC32 if needed #498

Closed benoit-pierre closed 1 year ago

benoit-pierre commented 1 year ago

It's only used by the caching, so only calculate it on full document load (i.e. not metadata only).

Resulting speedup during metadata extraction:

simulator on my laptop (1068 files): ~9% speedup, average ~12 → ~10 ms, minimum 5 → 5 ms, maximum 486 → 319 ms
on my Kindle Paperwhite (6th generation, 956 files): ×2 speedup: average 225 → 107 ms, minimum 84 → 84 ms, maximum 4862 → 400 ms

This change is