diff --git a/src/content/quickstart/index.html b/src/content/quickstart/index.html index 20319332f..5d33c389c 100644 --- a/src/content/quickstart/index.html +++ b/src/content/quickstart/index.html @@ -39,11 +39,11 @@
dotnet add package AngouriMath.FSharp
- The current release is 2.1.0, and both it and 2.0.0 change answers that earlier versions
+ The current release is 2.6.0, and every 2.x release changes answers that earlier versions
got wrong. Whichever version you are coming from, read
- BREAKING-CHANGES.md
+ BREAKING-CHANGES.md
first: it lists every input whose result is now different, with the old value and the new one,
- under a heading per release. 2.1.0 is a drop-in replacement for 2.0.0 — the assembly version is
+ under a heading per release. 2.6.0 is a drop-in replacement for 2.5.0 — the assembly version is
pinned at 2.0.0.0 for the whole of 2.x.
diff --git a/src/content/whatsnew/index.html b/src/content/whatsnew/index.html index 5ef8f61fe..ddf1bf7ed 100644 --- a/src/content/whatsnew/index.html +++ b/src/content/whatsnew/index.html @@ -34,7 +34,58 @@
+ A release about the integrator. Most of its pull requests move integration forward in every chapter
+ of Rubi's integration suite, from rational functions to the special functions, and each new route
+ checks its answer by differentiating it back. Of the 48,305 problems in that suite whose answers use
+ only functions AngouriMath has, 2.6.0 solves all 48,305 and none wrongly; 2.5.0 solved 7,580.
+ Statements became data: quantifiers, congruences,
+ divisibility, set relations and closed forms of sums. Numerical evaluation was rebuilt on
+ fixed-point series and interval arithmetic, and the test harnesses moved into CI.
+ Every input whose result is now different is listed, with both values and why, in
+ BREAKING-CHANGES.md
+ under 2.6.0 — since 2.5.0. Read that before upgrading.
+ The one to read first: | is divisibility now, and was disjunction, so "2 | 6"
+ is 2 divides 6. Writing or keeps the old reading.
+ Assembly version stays 2.0.0.0, and nothing public was removed, so this is a drop-in
+ replacement for 2.5.0. The full notes are on the 2.6.0 release.
+
+ The rewrite graph, tier 2 of the Math OS roadmap: an e-graph in the kernel, and every registered rule set running as data through an e-matcher. + The full notes are on the 2.5.0 release, and every changed answer is in + BREAKING-CHANGES.md. +
++ Tier 1 of the Math OS roadmap: rewrite rules written out as data, a pattern and a replacement. + The full notes are on the 2.4.0 release, and every changed answer is in + BREAKING-CHANGES.md. +
++ A correctness release measured on an outside corpus: more of Rubi's integration suite answered, and none of it wrong. + The full notes are on the 2.3.0 release, and every changed answer is in + BREAKING-CHANGES.md. +
++ A real polynomial layer, and a written specification of what canonical form means here. + The full notes are on the 2.2.0 release, and every changed answer is in + BREAKING-CHANGES.md. +
+A correctness release. Almost everything below is a wrong answer becoming a right one, and most of it was found by harnesses rather than reported — boundary points where a rule's assumption fails,