[koenvanderveen]

Summary

When a new account logs in, syft creates /SyftBox and the personal folder /SyftBox/{version}#{email}, but it didn't cache their ids. Later lookups searched Drive again. Now the id is cached when the folder is created, the same way the archive folder already does it.

Changes

• get_syftbox_folder_id and get_personal_syftbox_folder_id cache the id whether they find the folder or create it.
• The personal-folder search moves into get_personal_syftbox_folder_id_from_drive, to match get_syftbox_folder_id_from_drive.

Drive calls on a new account:

	Before	After
Login	3 searches	2 searches
First personal-folder lookup after login	1 search	0

Before this change, /SyftBox was only cached by accident: creating the personal folder searched for /SyftBox a second time. On real Drive, a search right after a create can miss the new folder because Drive's search index lags. That could create a second /SyftBox, and a second personal folder makes the next login fail with "Found 2 compatible folders". Caching the created id avoids that second search. I haven't seen this happen; it's a possibility, not an observed bug.

Testing

• New unit test: test_login_on_new_account_searches_each_folder_once counts Drive calls during login on a new mock account. It fails without the fix (4 searches) and passes with it (2).
• Unit, syft-rds, syft-datasets, syft-enclave, syft-job, migrations and mock-mode integration suites pass. Pre-commit is clean.

Asana task