Every test in the repository that needs to control time does so exclusively via env.ledger().set_timestamp(...) — grepping the entire test suite for any reference to the ledger's sequence number (as opposed to its timestamp), or any call to env.ledger().with_ledger_info(...), returns zero matches. Soroban's storage TTL and archival mechanics (relevant to the persistent-storage-with-TTL-bumps work already tracked in issue #236) are keyed off ledger sequence numbers, not timestamps, so the current test suite has no established pattern or helper for advancing sequence numbers at all. This means once #236's TTL-extension work lands, its tests will have to invent this test infrastructure from scratch rather than building on an existing convention. Consider adding at least one test today that exercises env.ledger().with_ledger_info(...) to advance the sequence number, establishing the pattern ahead of the TTL work.
Every test in the repository that needs to control time does so exclusively via
env.ledger().set_timestamp(...)— grepping the entire test suite for any reference to the ledger's sequence number (as opposed to its timestamp), or any call toenv.ledger().with_ledger_info(...), returns zero matches. Soroban's storage TTL and archival mechanics (relevant to the persistent-storage-with-TTL-bumps work already tracked in issue #236) are keyed off ledger sequence numbers, not timestamps, so the current test suite has no established pattern or helper for advancing sequence numbers at all. This means once #236's TTL-extension work lands, its tests will have to invent this test infrastructure from scratch rather than building on an existing convention. Consider adding at least one test today that exercisesenv.ledger().with_ledger_info(...)to advance the sequence number, establishing the pattern ahead of the TTL work.