Skip to content

v0.1.1

Choose a tag to compare

@skearnes skearnes released this 31 Aug 00:40
· 16 commits to main since this release
05fe1a6

No compatibility change since v0.1.0 — every path a v0.1.0 consumer used still resolves here. The corpus itself grew substantially over the intervening four years, and data/ now holds 550 .pb.gz and 53 .parquet datasets: the Parquet files were added alongside the originals rather than replacing them.

This is the last release carrying the legacy .pb.gz files. v0.2.0 removes them.

Schema changes since v0.1.0

The corpus is validated against ord-schema 0.8.3, up from v0.3.0 at v0.1.0. Comparing the two definitions by field number — what the wire format actually keys on — the change is additive: reaction.proto goes from 255 to 266 fields and 321 to 341 enum values with no field removed, renumbered, or retyped, and dataset.proto is unchanged. Several nested types were renamed or hoisted (TemperatureConditions.Measurement → TemperatureMeasurement, ProductCompound.Texture → Texture, and the per-message IdentifierType/WorkupType-style enums), each keeping its numbering exactly.

Two symbols were renamed in place, both keeping their number:

Was Is Number
ReactionIdentifier.IdentifierType.NAME REACTION_TYPE 5
Compound.Source.id catalog_id 2

Neither changes what stored data means — value 5 carries the same documented meaning it always did. They are source-level breaks in ord_schema itself: code referencing the old names needs updating when the library is upgraded, independent of which ord-data release it reads.

Reaction data lives in Git LFS, so the auto-generated source archives below contain pointer files, not reactions. To get the data:

git clone https://github.com/open-reaction-database/ord-data.git
# or, without git history:
python scripts/download_from_huggingface.py

LFS reads are served by the Hugging Face mirror.

What's Changed

New Contributors

Full Changelog: v0.1.0...v0.1.1

🤖 Generated with Claude Code