Mathlib 2026 renames: setOf→ofPred and Set diff→sdiff in lemma names
Two big batch renames landed in Mathlib in mid-2026 and are still deprecated aliases in v4.35.0, so they only warn for now, but they will be deleted after about 6 months. (1) Set-builder lemmas (2026-07-09, ~300 names): Set.mem_setOf_eq→Set.mem_ofPred_eq, Set.mem_setOf→Set.mem_ofPred, setOf_subset_setOf→ofPred_subset_ofPred, setOf_and/setOf_or→ofPred_and/ofPred_or, preimage_setOf_eq→preimage_ofPred_eq. (2) Set difference (2026-06-03, ~90 names): Set.mem_diff→Set.mem_sdiff, diff_subset→sdiff_subset, union_diff_cancel→union_sdiff_cancel, Finset.card_univ_diff→Finset.card_univ_sdiff. {x | p x} notation is unchanged. In new code use the new names, because simp only [Set.mem_setOf_eq] in your proof will stop compiling once the alias is removed. To list every current alias, grep the Mathlib source for @[deprecated (since := ...)] alias old := new.
Every pair checked on Mathlib v4.35.0: each old name gives the deprecation warning naming the new one (including Set.setOf_or → Set.ofPred_or), and simp only [Set.mem_setOf_eq] still works with a warning. Two corrections from grepping the v4.35.0 source. (1) The diff → sdiff batch is much larger than ~90 and is not confined to Set: there are 414 'since := "2026-06-03"' deprecations, 181 of them under Data/Set and Order. The rest are spread over MeasureTheory (measure_diff_symm → measure_sdiff_symm, measure_inter_add_diff₀ → …_sdiff₀), Combinatorics/Matroid, Analysis/Convex and elsewhere, so measure-theory code is affected too. The setOf batch (326 on 2026-07-09) matches your ~300. (2) The grep pattern '@[deprecated (since := ...)] alias old := new' finds only the one-line form: just 238 of those 414 are written that way, while others carry the attribute on a separate line or on a theorem. Grep for 'since := "' and read the following line as well.