[AshutoshMore142k4]

Follow-up to #257. That PR was merged with one review finding still open: a retained value with an explicit TTL change was treated as already held.

Problem

gddy dns set --type A --name www --data 1.2.3.4 --ttl 7200 on a zone that already holds 1.2.3.4 at 3600:

1. plan_set pairs the value with its existing record (same_content ignores ttl, on purpose).
2. apply_replace sees the explicit TTL change and creates the new record before deleting the old one.
3. v3 identifies a record by its content, not its TTL (the premise of fix(dns): pair retained values before position in the set plan #257), so that create is an exact duplicate: DUPLICATE_RECORD. The old record stays at 3600 and the command fails.

This predates #257 (positional pairing paired the same two records), but the TTL-only case is the one #257's review flagged and it was never addressed.

Fix

In apply_replace, a kept value (same_content) is now one of two things:

• no explicit --ttl, or the same TTL as stored: no-op, as before.
• explicit --ttl that differs: retime_record deletes the old record, creates it with the new TTL, and puts the old record back unchanged if the create fails. Only if that restore also fails is a cleanup outcome added saying the record is gone and how to re-create it.

Deleting first opens a short window where the value is missing. With only create/delete on the record endpoint (no in-place update, see #136) that cannot be avoided for a TTL-only change, hence the restore.

Values that are not kept still go through create-then-delete, unchanged.

Tests

retime_* in rust/src/dns/set/write_tests.rs:

• delete fails: no create is attempted (proves delete runs first)
• create fails: old record is restored (POST with the old TTL), failure reported
• create and restore both fail: cleanup outcome

The existing apply_replace_is_not_a_no_op_when_ttl_explicitly_changes and ..._is_still_a_no_op_when_ttl_is_omitted... still pass.

write.rs would have passed the 1000-line limit, so its test module moves to write_tests.rs (same #[path] pattern as hosting/client.rs). That part of the diff is a dedent only; the only changed lines are rustfmt re-wrapping a few json_body calls.

Checks

cargo check --workspace, cargo clippy --workspace -- -D warnings, cargo test --workspace, cargo fmt --check and rust/scripts/check-module-size.sh all pass.

Not verified against the live API: the create-after-delete order is exercised with mock servers only.

[Copilot]

Copilot was unable to review this pull request because the user who requested the review has insufficient remaining quota. Try requesting a review at a lower effort level.

An organization owner can choose the repository organization as the billing source for reviews requested by members with a paid Copilot license. Learn about Copilot code review billing.

[jpage-godaddy]

Thanks for the contribution, @AshutoshMore142k4