[ccomb]

A ranking of 30 yoghurts under a normalized and weighted single score (climate at 21 %, water at 8.5 %, and so on) rests on value choices, not on a measurement. ISO 14044 does not allow a weighted score in a comparison disclosed to the public, and nothing said so: "yoghurt A has 2.7 times less impact than yoghurt B" read straight off the ranking.

Now:

• a scoring set says whether it weighs (list_scoring_sets, the scoring sets route);
• a ranking under such a set says it once, as reading_notice (score_activities) or readingNotice (the batch route). The ranking is still given.

A set weighs when one of its scores takes a category otherwise than once (its weight, its coefficient in the score, or its coefficient in the row the score reads), adds up categories it normalizes, or is no sum at all. A normalized category scored alone does not weigh, nor do damages in one unit added at one: a variable without a weight counts for nothing in this engine, so weights of one only add categories up. Categories in different units added at one are not caught; that is said in the code and in the CHANGELOG.

Left out on purpose: the score of one activity and its breakdowns rank nothing; compare_impacts scores one activity on two sides; compare_activities compares inventories.

Wire revision 105. pyvolca reads BatchScores.reading_notice; weighs reaches the dictionaries scoring_sets() returns.

Checks: cabal build all --ghc-options=-Werror, cabal test lca-tests, fourmolu, hlint (with the partials gate), and pytest on pyvolca.