On this page
What changed in each release
This page is for someone who loaded one release and wants to know why rows moved in the next. Each release lists what changed and why.
To apply the changes to a database, use the migration files published with each release; see Which output do I need?. They list every identifier that changed. This page says why, grouped by what you would notice.
1.1.0 — 2026-10-21
Published on Zenodo.
Built from the same five lists as 1.0.0: ecoinvent 3.12, ecoinvent 3.8, BAFU 2026 v1, Stepwise 2006 1.09 and AGRIBALYSE 3.2. A count below comes from the full build its pull request was measured on. That pull request names the build.
1.0.0 was built on 2026-09-20 from commit 50543d9, with uncommitted changes.
That commit is no longer in the repository, so the changes made between that
build and 22 September 2026 are not itemised here. The migration from 1.0.0
covers them along with everything else.
The migration from 1.0.0
1.0.0__1.1.0.zip moves a database that loaded 1.0.0 to 1.1.0. On the trial
build of 1.1.0 (commit 15d733d, run 20261021T2030219167640000) it holds:
| Created | Updated | Replaced | Deleted | Left for a person | |
|---|---|---|---|---|---|
| Substances | 2 | 143 | 3 | 0 | 0 |
| Flows | 31 | 1,729 | 22 | 0 | 0 |
| Impact categories | 0 | 44 | 0 | 0 | 0 |
| Factors | 139 | 345,997 | 0 | 0 | 0 |
Each replaced substance or flow is also listed under delete in its file, so
that it leaves your list of flows. It is counted once here, as replaced. The
44 updated categories and 345,997 updated factors are the renaming described
in the next section. Once the migration is applied, a database
holds 1.1.0's 8,159 substances, 97,827 flows, 138 impact categories and
727,074 factors, and every one of 1.0.0's 97,818 flows in an inventory points
at a flow 1.1.0 publishes. That was checked by applying the files to 1.0.0
with randonneur and comparing the result with 1.1.0, table by table. Nothing
is left unresolved.
If you hold 1.0.0
- Our own impact categories have new identifiers. 1.0.0 published our
implementations of EF 3.1 and Stepwise 2006 under the name
consensus-flow-list. They are now published asbrightway-flows. A category's identifier is made from a name that includes that word, so all 44 categories changed identifier. The migration renames each category and moves its factors with it, all 345,997 of them. It does not delete them and recreate them (#410). - The files now say which build made them. In
harmonised-flows-simple.json.gzit isowl:versionInfo, and inlcia-factors.json.gzit isstats. Each names the build run and the commit (#414). 1.0.0's files name neither. - The release carries its own record. Besides the usual files, the
Zenodo record holds this
release's snapshot,
release-snapshot.json.gz, and1.0.0__1.1.0.zip, the migration from 1.0.0 to 1.1.0.
Substances and their names
- Ions are spelled one way.
Lead IIandLead(II)are nowLead(2+),ChlorideisChloride(1-),Copper, IonisCopper, ionandCesium(1+)isCaesium(1+). All 51 ion substances follow one standard, and 14 of them changed label. Every other spelling stays as a synonym on the substances and on all 727 of their flows, so a search forLead IIstill finds lead(2+) (#402). - Rhenium(2+) no longer carries rhenium's numbers. EF 3.1's 13
rhenium(2+)emission flows shipped the metal's CAS number, 7440-15-5, and its EC number. Both have been removed. Four ions are now typed as cations rather than as imprecise mixtures: rhenium(2+), palladium(2+), rhodium(3+) and chromium(4+). Rhenium and rhenium(2+) have new substance identifiers. Their flows and factors are unchanged (#406). - Fenpropimorph is one substance. EF 3.1 published it twice:
- 13 flows under 67564-91-4, 11 of them characterised;
- 12 flows named
fenpropimorphunder the older number 67306-03-0, none of them characterised.
BAFU's one fenpropimorph row and AGRIBALYSE's five carry the older number,
so all six reached flows with no factors. The two substances are now one,
under 67564-91-4, and those six rows reach characterised flows. The 12 flows
that had no factors are deprecated onto their characterised twins.
67306-03-0 is no longer published, so look the substance up by 67564-91-4
(#424).
- 2,3,4,5-tetrachlorobenzoyl chloride keeps the space in its name. Its
IUPAC name used to depend on the order in which one build happened to read
two spellings. 1.0.0 published …benzoylchloride, without the space
(#419).
Where a source row lands
- ecoinvent 3.8's bare metals are the element. ecoinvent 3.8 ships
Leadto urban air under 7439-92-1. From 3.9.1, the same identifier isLead II. 1.0.0 followed the identifier and published 3.8's row on lead(2+). Now each of 299 such rows reaches the element its CAS number names: lead, copper, molybdenum and others. - 41 of these rows now reach a characterised flow where before they reached an uncharacterised one.
- 28 new flows appear, none of them characterised.
- 6 ion flows at aircraft cruise height are gone; only these rows reached them.
The full table is on the ecoinvent 3.8 page
(#400).
- AGRIBALYSE's generic arsenic ion is generic. All 13 of its
Arsenic, ion rows shipped pentavalent arsenic's number, 17428-41-0, so they
reached arsenic(5+). They now reach Arsenic, ion, where ecoinvent's and
BAFU's rows already were. Its four Arsenic (V) rows stay on arsenic(5+)
(#422).
- AGRIBALYSE's coal-mine off-gas is no longer natural gas. Its three
Gas, mine, off-gas, process, coal mining rows carried natural gas's
number, 8006-14-2, so they reached Natural Gas. They now reach the
coal-mine off-gas flow
(#409).
Units
- AGRIBALYSE's 26 rows in another unit are converted. Each of these rows reaches a flow published in a different unit, and none of them carried a conversion. For example, 16.7 million m³ of natural gas was published as 16.7 million MJ, about 35 times too little. These rows are now converted as follows (#409):
- fuels at AGRIBALYSE's own heating values: natural gas at 34.5 MJ/m³, hard coal at 18.2 MJ/kg;
- uranium at 560,000 MJ/kg;
- radionuclides given by mass, at their specific activity;
- water given by mass, at 0.001 m³/kg.
Two stable isotopes are accepted with no number.
- BAFU rows on a flow in another unit carry the factor. BAFU often ships a
row in becquerels that reaches a flow published in kilobecquerels. Such rows
now carry the factor from the unit table, 0.001 in that case, and say where
it came from:
- 162 rows that reach an existing flow
(#399);
- 56 rows whose flow they themselves created
(#416).
- The unit-mismatch flag means one thing. In consensus-flows.sqlite3,
has_unit_mismatch used to stay set on a matched row even when a curated
factor said by how much its unit differs, such as 9.41 MJ per kilogram. It
is now cleared there, as it already was on a row that created its flow
(#420).
Looking up a flow
These change the answers flow lookup gives. They do not change the published list.
- A conversion is reported in full. If you send a BAFU row in becquerels, the lookup returns the target in kilobecquerels together with the 0.001 that converts it. It shows each step: your unit to the unit the list shipped, then any multiplier the list recorded, then the target's unit. A conversion nobody recorded comes back empty, never as 1.0 (#387).
- BAFU's
resources / unspecifiednames no medium. BAFU'sCarbon dioxide, in airrow is filed underresources / unspecified. A query for it that named BAFU was read as a resource in the ground, and so it found nothing. That compartment is now read as a resource with no medium. The eight rows the old reading was right about are ruled one by one (#390). - A row is found under the name the vendor shipped. BAFU's
t-Butylaminesent without its CAS number used to find nothing, because the build had recorded it under the enriched nametert-Butylamine. Recorded decisions are now found under both names (#423).
Not visible in the published files
These changes are to how builds are checked, not to anything a build publishes: - the recorded expectations and their baseline (#391, #392, #397); - a test of a pinned category identifier (#405); - the source table in What was changed in each source (#411); - the tool that compares two builds, which now shows a change to a published conversion factor (#421).
1.0.0 — 2026-09-20
The first release, published on Zenodo. It merged ecoinvent 3.12, ecoinvent 3.8, BAFU 2026 v1, Stepwise 2006 1.09 and AGRIBALYSE 3.2 onto EF 3.1. It holds:
- 97,818 live flows;
- 2,676 redirects;
- 726,935 factors.
It cannot be rebuilt from source. Its snapshot was recorded from the database published with it; see Running a transform.