58.10.1.1 Querying Retained Source🔗ℹ

58.10.1.2 Planning String Correspondences🔗ℹ

When both appearance keywords are supplied, the part keeps its ordinary route, but its changed appearance completes when that route first reaches the current x-coordinate of appearance-complete-at-x. The duration is a fraction of the enclosing transition: the old and new appearances cross-fade only in the interval immediately preceding that deadline. This is useful for a moving + that should become - exactly as it passes an equality sign.

58.10.1.3 Part Routes and Copies🔗ℹ

58.10.1.4 Generated SVG Fragments🔗ℹ

58.10.1.5 Part and Glyph Transitions🔗ℹ

Exact matches render as one rigid SVG group that moves with its local transform. Changed explicit matches are moving cross-fades, and unmatched fragments use the normal fade-out/fade-in behavior. This operation does not infer algebraic equivalence, parse TeX tokens, choose paths/arcs for the movement, or morph glyph outlines.

This matches and moves whole rendered glyph leaves. 'fade is the default: changed matches use the ordinary moving cross-fade. With #:changed-mode 'morph, a changed matched pair instead interpolates its outline only when both cropped dvisvgm SVG fragments expand to one identically painted path whose positive-length contours are all closed and compatible in count. Animate globally pairs those destination contours with the source, phase-aligns them without reversing their traversal, normalizes the resulting paths to compatible cubic segments, and uses that path geometry only for interior frames; the ordinary tagged SVG fragments remain exact endpoints. Glyphs with multiple independently painted paths, open contours, incompatible contour topology, changed paint, or unsupported geometry safely fall back to the moving cross-fade.

This operation does not derive a mathematical operation, identify TeX characters or terms, perform semantic grouping, or infer which changed glyphs should be paired.

58.10.1.6 Anchored Rewrites🔗ℹ

When scene-play compiles the request, Animate translates the complete destination layout so the destination anchor coincides with the corresponding part in the current source formula. Consequently, a sequence of rewrites keeps the anchor fixed even when the formula values passed as earlier templates were constructed at their own default positions. The translation preserves the target formula’s TeX spacing and baselines.

Each stationary entry names an additional matched pair: a symbol means the same source and destination part name, while a formula-part-match permits different names. The pair is made explicit, and at clip compilation the destination fragment receives the current source fragment’s exact affine transform. Thus several selected terms can remain fixed even if the rest of the destination layout moves or reflows. This is an explicit presentation choice; it does not infer which terms should remain still or maintain a general layout constraint between them.

The remaining keywords have the same meaning as in transform-matching-parts: explicit matches take priority, routes and copies select intentional term motion, and 'fade-transform cross-fades remaining unmatched parts while moving them. Like the lower-level operation, this is whole-fragment correspondence rather than TeX parsing or glyph-outline morphing.

58.10.1.7 Derivation Steps🔗ℹ

anchor defaults to #f, which means that the derivation’s shared anchor is used. A step can override it with a same-name symbol or an explicit formula-part-match. stationary has the same meaning as in rewrite-formula and makes additional matched parts fixed for this one step. This data does not claim that the rewrite is algebraically valid; it records the author’s chosen presentation.

When any step has an explanation, supply explanation-position. The builder creates plain text with explanation-id, which must be absent from the initial scene. Later explanations replace only that generated Visual. The final explanation remains visible unless a later step omits it.

This is immutable convenience syntax over existing scene and formula APIs. It does not parse TeX, infer operations, prove a derivation, choose matches/routes, or automatically lay out the explanation.