README.md:241-242 says:
Features. Nothing writes CIM, and a non-conformant CGMES file is worse than none. Time series are read whole rather than streamed.
Both halves are now false, and have been since 26 July 2026:
- CIM is written.
crates/gridwright-io/src/cgmes_write.rs (d6a2e71, f27d5b4), with 14 round-trip and conformance tests in crates/gridwright-io/tests/cgmes_write.rs — including what_cim_cannot_hold_is_reported, which is the honest answer to the conformance worry the sentence is expressing.
- Time series are streamed.
9bf23e4 "Bound memory while reading a large time series", pinned by a_year_of_hourly_data_does_not_have_to_be_resident_all_at_once in crates/gridwright-io/tests/streaming.rs.
The README was rewritten on 28 July (42bd7c1), two days after both landed, and carried the stale claim through. Worth fixing promptly: this is the repository's public description of its own limits, and it is currently underselling the reader by two formats.
TODO.md:684 and TODO.md:691 carry the same stale framing and should be ticked at the same time.
README.md:241-242says:Both halves are now false, and have been since 26 July 2026:
crates/gridwright-io/src/cgmes_write.rs(d6a2e71,f27d5b4), with 14 round-trip and conformance tests incrates/gridwright-io/tests/cgmes_write.rs— includingwhat_cim_cannot_hold_is_reported, which is the honest answer to the conformance worry the sentence is expressing.9bf23e4"Bound memory while reading a large time series", pinned bya_year_of_hourly_data_does_not_have_to_be_resident_all_at_onceincrates/gridwright-io/tests/streaming.rs.The README was rewritten on 28 July (
42bd7c1), two days after both landed, and carried the stale claim through. Worth fixing promptly: this is the repository's public description of its own limits, and it is currently underselling the reader by two formats.TODO.md:684andTODO.md:691carry the same stale framing and should be ticked at the same time.