pospi
commented
2 years ago on various mechanisms for cross-DNA validations:
(via @pdaoust)
- FWIU cross-cell
must_get_*is in the works 👍🏼 I don't see it on the roadmap, but I wonder if it's rolled into the 'Holochain Resource locators' milestone? - signed claims with expiry times would certainly work deterministically -- mimicking TLS certs but without the convenience of revocation lists
- locking/semaphores would work in coordinator zomes as a precaution against doing something stupid, but (at least the way I'm picturing them working) they would work on the observed state of a particular source chain that's doing the work of maintaining the state of locks, which might make it sorta non-deterministic functions. This could work in validation functions though, maybe, if a "this was valid as-at a certain point in source chain X" assurance was sufficient. (And if there were a good way of ensuring the accuracy of timestamps, or alternatively that the timestamps didn't matter so much.)
https://lab.allmende.io/valueflows/valueflows/-/wikis/Application-specification-guidance
Create / update logic should mostly validate as expected, though a few new protocol additions since the first implementation have yet to be caught up on.
The presence of associated records is rarely validated upon creation. There could be some novel ways of implementing these checks by providing self-signed receipts from the agent of a foreign DNA in order to prove that other record(s) have been written when committing an index update.
Deletions are mostly not validated at all. Some affordance of DNA dependencies by Holochain Core might be necessary in order to implement such features.