Hashcode-cache (cinbox-temp) causes false-positives when resetting files.

[pjotrek-b]

Currently, the CInbox calculates hashcodes of ITEM contents only once (using the TashHashGenerate), and stores them in CINBOX_TEMP folder.

But, in the following cases that's a problem, when a valid reference hashcode-manifest is used for integrity checking:

1. An ITEM file was actually corrupt, and correctly detected using HASH_SEARCH. The CInbox Item goes to Error.
2. An ITEM file was not corrupt, but a wrong copy/version was used.

Once the "problematic" file is replaced by a valid copy, and the ITEM is reset into Todo, the cinbox-temp contains the hash from the previous run (=from the broken file), so it throws a false-positive Error.

In cases where the CInbox is running on backend servers, and the CInbox operator has no write access to CINBOX_TEMP, the operator cannot resolve this issue.

[pjotrek-b]

The hash-cache was originally intended and implemented for two reasons:

a. To avoid recalculating hashes on re-reruns.
b. To be able to re-use the hashes in later tasks. Like for HashSearch, HashValidate and HashOutput.

So if I moved the HashGenerate task /after/ HashSearch, this issue should be resolved, as there is no "old" hashcode for HashSearch to use.

It may be however, that HashSearch currently expects the hash-cache to be filled.
I'll check.

[pjotrek-b]

Maybe a more declarative solution would be to add a new config option for HashSearch to force fresh calcuation of the hashes and bypass any cache.