In https://github.com/MicrosoftEdge/MSEdgeExplainers/blob/master/CacheAPIResponseMetadata/explainer.md, a guarantee that lastAccessedAt and retrievalCount are exact might cause problems similar to atime in unix. (Or might not; this is just a worry.) The atime problems were solved by relatime, which makes the stored data less exact, and lazytime, which might lose updates if the system crashes.
I see a note in https://github.com/MicrosoftEdge/MSEdgeExplainers/blob/master/CacheAPIResponseMetadata/explainer.md#privacy-considerations that lastAccessedAt might have similar-to the relatime semantics anyway: that should be in its initial description, and you should consider guaranteeing that it's at least as new as cachedAt.
In https://github.com/MicrosoftEdge/MSEdgeExplainers/blob/master/CacheAPIResponseMetadata/explainer.md, a guarantee that
lastAccessedAtandretrievalCountare exact might cause problems similar toatimein unix. (Or might not; this is just a worry.) Theatimeproblems were solved byrelatime, which makes the stored data less exact, andlazytime, which might lose updates if the system crashes.I see a note in https://github.com/MicrosoftEdge/MSEdgeExplainers/blob/master/CacheAPIResponseMetadata/explainer.md#privacy-considerations that
lastAccessedAtmight have similar-to therelatimesemantics anyway: that should be in its initial description, and you should consider guaranteeing that it's at least as new ascachedAt.