Regina - Moving from 5.x to 6.0

Over time, Regina makes changes and improvements to its C++ and Python interfaces. This page explains how you will need to update your C++ code and/or Python scripts to work with Regina 6.0 and later.

Now that Regina 7.0 is available, you should also read the separate pages on:

• moving from Regina 6.x to 7.0 (December 2021) – this was an enormous update that made sweeping changes throughout the API;
• recent changes since Regina 7.0.

If you have older code (from 2016 or earlier), you can also read:

• moving from Regina 4.x to 5.0 (September 2016).

Changes from Regina 5.x to 6.0

There are some general rules that cover most of the changes:

• Classes of the form NFoo have been renamed to Foo (e.g., NAbelianGroup is now AbelianGroup).
• Classes of the form NFooList have been renamed to Foos (e.g., NAngleStructureList is now AngleStructures).
• For classes related to triangulations (i.e., triangulations, vertices/edges/triangles/etc., simplices, components, boundary components, and isomorphisms):
  • 3-D classes NFoo have been renamed to Foo<3> (e.g., NTriangulation is now Triangulation<3> in C++, or Triangulation3 in Python).
  • 2-D and 4-D classes Dim2Foo and Dim4Foo have been renamed to Foo<2> and Foo<4> (e.g., Dim2Edge is now Edge<2> in C++, or Edge2 in Python).
  • All of these classes are now included by a single "unified" C++ header for each dimension: triangulation/dim2.h, triangulation/dim3.h, triangulation/dim4.h, or triangulation/generic.h for dimensions ≥ 5.
• Classes of the form NNormalSurfaceVectorFoo and NNormalHypersurfaceVectorFoo have been renamed to NSVectorFoo and HSVectorFoo respectively.
• All C++ headers have been renamed in the same way as their classes, unless otherwise specified (e.g., nabeliangroup.h is now abeliangroup.h).

Beyond these general rules, there are some additional changes and/or special cases. These are listed in the table below, along with suggested replacements that you can use instead. Any changes in the behaviour of these routines or classes are noted in the comments column.

Deprecated code	Replacement	Comments
Headers
approx.h	—	These functions for managing round-off error are very coarse, and will be removed because they should not be used.
engine.h, output.h	core/engine.h, core/output.h
nbooleans.h	boolset.h
nlayeredsurfacebundle.h	layeredtorusbundle.h
nmatrixint.h	matrix.h
nperm2.h, nperm3.h, nperm4.h, nperm5.h	perm.h
packetlistener.h	packet.h
sfcombination.h, sfproperties.h	surfacefilter.h
Classes
Dim2EdgePairing, NFacePairing, Dim4FacetPairing	FacetPairing<2>, FacetPairing<3>, FacetPairing<4>
Dim2ExampleTriangulation, NExampleTriangulation, Dim4ExampleTriangulation	Example<2>, Example<3>, Example<4>
Dim2TriangleEdge, NTetFace, Dim4PentFacet	FacetSpec<2>, FacetSpec<3>, FacetSpec<4>
MatrixIntDomain, MatrixRing	Matrix	These have been merged into the main Matrix class, and the extra functionality that they provided is now handled using template arguments.
NExampleSnapPeaTriangulation	ExampleSnapPea
NNormalHypersurfaceVectorMirrored	—
NSurfaceSubset	—
NPropertyBase	—
Types
TuraevViroAlg	Algorithm
UseDim2GluingPerms, UseGluingPerms, UseDim4GluingPerms	GluingPermSearcher<2>::Use, GluingPermSearcher<3>::Use, GluingPermSearcher<4>::Use
Member Routines
Matrix::multiplyAs()	operator *=	Now that the Matrix class is unified, there is no need for the type-specific multiplyAs().
NSVectorMirrored copy constructor	—
Triangulation<2>::oneThreeMove(), Triangulation<2>::twoTwoMove(), Triangulation<2>::threeOneMove()	Triangulation<2>::pachner()
Triangulation<3>::drillEdge()	Triangulation<3>::pinchEdge()	In all cases that matter, this performs the same topological operation with vastly fewer tetrahedra.
Triangulation<3>::knowsSplittingSurface()	—	The routine hasSplittingSurface() is now fast, and so there is no need to pre-query whether the result is already known.
Triangulation<3>::oneFourMove(), Triangulation<3>::twoThreeMove(), Triangulation<3>::threeTwoMove(), Triangulation<3>::fourOneMove()	Triangulation<3>::pachner()
Triangulation<4>::oneFiveMove(), Triangulation<4>::twoFourMove(), Triangulation<4>::threeThreeMove(), Triangulation<4>::fourTwoMove(), Triangulation<4>::fiveOneMove()	Triangulation<4>::pachner()
Global Routines
isZero(), isNonZero(), isPositive(), isNegative(), isNonPositive(), isNonNegative()	—	These methods for managing round-off error are very coarse, and will be removed because they should not be used.
Constants
Matrix::zero, Matrix::one	0, 1
PACKET_DIM2TRIANGULATION, PACKET_TRIANGULATION, PACKET_DIM4TRIANGULATION	PACKET_TRIANGULATION2, PACKET_TRIANGULATION3, PACKET_TRIANGULATION4
PACKET_NORMALSURFACELIST, PACKET_NORMALHYPERSURFACELIST, PACKET_ANGLESTRUCTURELIST	PACKET_NORMALSURFACES, PACKET_NORMALHYPERSURFACES, PACKET_ANGLESTRUCTURES
TV_DEFAULT, TV_BACKTRACK, TV_TREEWIDTH, TV_NAIVE	ALG_DEFAULT, ALG_BACKTRACK, ALG_TREEWIDTH, ALG_NAIVE	The corresponding enum type has changed from TuraevViroAlg to the more general Algorithm.
Command-Line Utilities
All MPI utilities (e.g., tricensus-mpi)	—	These have been phased out in favour of other approaches to large-scale parallelisation. See the tricensus-mpi manual page for suggestions.

Back to main page ...