diff --git a/doc/VERSIONS.txt b/doc/VERSIONS.txt
index cf6998a5b9dace105f5dd968d6889cdc563e65f9..cc6c67d18c7d2448f23f384432f2d28b91709d9f 100644
--- a/doc/VERSIONS.txt
+++ b/doc/VERSIONS.txt
@@ -1,4 +1,7 @@
-$Id: VERSIONS.txt,v 1.42 2009-03-18 20:29:26 geuzaine Exp $
+$Id: VERSIONS.txt,v 1.43 2009-03-30 08:19:28 geuzaine Exp $
+
+2.3.2 (?): optionally copy transfinite mesh contraints during geometry
+transformations.
 
 2.3.1 (Mar 18, 2009): removed GSL dependency (Gmsh now simply uses
 Blas and Lapack); new per-window visibility; added support for
diff --git a/doc/texinfo/gmsh.texi b/doc/texinfo/gmsh.texi
index 4a5f740905c4d55f4cc5ddb9b5a93e847cb5b779..f0c2f0f5c635832035bc1a61c88bce56d8ea093d 100644
--- a/doc/texinfo/gmsh.texi
+++ b/doc/texinfo/gmsh.texi
@@ -2969,12 +2969,12 @@ interpolation matrices used for high-order adaptive visualization.
 Let us assume that the approximation of the view's value over an element
 is written as a linear combination of @var{d} basis functions
 @var{f}[@var{j}], @var{j}=0, ..., @var{d}-1 (the coefficients being
-stored in @var{list-of-values}). If @var{f}[@var{j}] = @var{p}[0]
-@var{F}[@var{j}][0] + @var{p}[1] @var{F}[@var{j}][1] + @var{p}[2]
-@var{F}[@var{j}][2] + ..., with @var{p}[@var{i}] =
+stored in @var{list-of-values}). Defining @var{f}[@var{j}] =
+Sum(@var{i}=0, ..., @var{d}-1) @var{p}[@var{i}]
+@var{F}[@var{j}][@var{i}], with @var{p}[@var{i}] =
 @var{u}^@var{P}[@var{i}][0] @var{v}^@var{P}[@var{i}][1]
 @var{w}^@var{P}[@var{i}][2] (@var{u}, @var{v} and @var{w} being the
-coordinates of the element's parameter space), then
+coordinates in the element's parameter space), then
 @var{val-coef-matrix} denotes the @var{d} x @var{d} matrix @var{F} and
 @var{val-exp-matrix} denotes the @var{d} x @var{3} matrix @var{P}.
 
@@ -2982,14 +2982,13 @@ In the same way, let us also assume that the coordinates @var{x},
 @var{y} and @var{z} of the element are obtained through a geometrical
 mapping from parameter space as a linear combination of @var{m} basis
 functions @var{g}[@var{j}], @var{j}=0, ..., @var{m}-1 (the coefficients
-being stored in @var{list-of-coords}).
-
-If @var{g}[@var{j}] = @var{q}[0] @var{G}[@var{j}][0] + @var{q}[1]
-@var{G}[@var{j}][1] + @var{q}[2] @var{G}[@var{j}][2] + ..., with
-@var{q}[@var{i}] = @var{u}^@var{Q}[@var{i}][0]
-@var{v}^@var{Q}[@var{i}][1] @var{w}^@var{Q}[@var{i}][2], then
-@var{val-coef-matrix} denotes the @var{m} x @var{m} matrix @var{G} and
-@var{val-exp-matrix} denotes the @var{m} x @var{3} matrix @var{Q}.
+being stored in @var{list-of-coords}). Defining @var{g}[@var{j}] =
+Sum(@var{i}=0, ..., @var{m}-1) @var{q}[@var{i}]
+@var{G}[@var{j}][@var{i}], with @var{q}[@var{i}] =
+@var{u}^@var{Q}[@var{i}][0] @var{v}^@var{Q}[@var{i}][1]
+@var{w}^@var{Q}[@var{i}][2], then @var{val-coef-matrix} denotes the
+@var{m} x @var{m} matrix @var{G} and @var{val-exp-matrix} denotes the
+@var{m} x @var{3} matrix @var{Q}.
 
 Here are for example the interpolation matrices for a first order
 quadrangle: