What about moving the vend_acct field onto the actual provider record? Then the EDI account should only include the username/password and the label could be something like Library-Baker & Taylor (all).
So that would mean that:
The fetcher would need to make the connection, if there are any files, compare if they already exist on that connection, if not pull them down. Then do the matching based on the account field on provider records, not the EDI account.
The pusher constructs the EDI message and pulls the vendacct from the EDI account too, it would have to pull it from the account field on the provider record. This seems better, since it wouldn't have to make the jump.
So theoretically, we'd have less EDI accounts, so less connections made.
Upgrade issues:
How would we stop a million old files from being pulled in since it's looking per EDI account what's been pulled in already? Should we put a deleted/inactive flag on EDI accounts?
https://bugs.launchpad.net/evergreen/+bug/1836908
What about moving the vend_acct field onto the actual provider record? Then the EDI account should only include the username/password and the label could be something like Library-Baker & Taylor (all).
So that would mean that:
So theoretically, we'd have less EDI accounts, so less connections made.
Upgrade issues: