adranwit
commented
1 year ago The cache fs system is designed to be used on cloud storage, where folder structure can be stored in just one cache file, from design perspective, it's both performance and cost optimization where you would avoid tons of class A and B operation call on the cloud, In the latest version the ignore option has been added, where you can exclude asset based on specified criteria
Currently, a lot of data source reads are cached; this may be useful for certain situations, especially if within the API there is support for copying the same data source multiple times.
However, this cache is too aggressive by default - it will not clear the cache until the process restart. There is not much precedent in infrastructural tools exposing a file-system level API that will inherently cache so aggressively. There may need to be a general system that requires checking if a cache is dirty, or expiring the cache within a reasonable scope.