Release Notices for Cactus 4.0 beta 3
=====================================
Sunday October 24th 1999


Released modules:
================

Cactus
CactusBase
CactusEinstein
CactusElliptic
CactusNet
CactusPUGH
CactusPUGHIO
CactusWave
external


Main Changes:
============
Initial implementation of general dimension grid functions
(CactusPUGH/PUGH)
Dynamic courant condition (CactusBase/Time)
Ghostzone generality (CactusPUGH/PUGH)
Faster compilation times (Cactus)


CVS Tag: Cactus_4_0_Beta_3
=======


==============================================================================


CHANGES FOR BETA 3
==================

Utils
=====

1) The perl processing part of the compilation process
now only replaces files which have actually changed,
so that compilation times are now much quicker (after
the first time).

2) The test suite checker doesn't report a failed test if
thorns needed are missing.

Util Thorns
===========

1) The names of boundary condition routines in
CactusBase/Boundary have been standardised and
renamed to make a distinction between groups and
variables. The old names remain, but will be removed
in Beta 4. See the documentation in this thorn for
details.

2) Thorn CactusBase/Time now contains a routine for
using a dynamic courant condition to set the timestep.
This is implemented in CactusEinstein/Einstein for
the relativity thorns ... see the documentation in
Time for details.

3) The standard PUGH driver now supports different
numbers of ghostzones in different directions.
Also, there is a now a parameter for setting
the number of grid points on each local processor.

4) The standard PUGH driver now supports 1D grid functions.

5) The IsoSurfacer now supports display from Linux machines.


Relativity Thorns
=================

1) Einstein now uses thorn CactusBase/Time to set the
timestep (previously Einstein set it itself). This
means that you now have to compile Time and add it
to your active thorns, and the parameter einstein::dtfac
should be changed to time::dtfac in parameter files.

2) Bugs in misner data and Brill Lindquist data have
been fixed


==============================================================================


KNOWN BUGS IN BETA 3
====================
(See Cactus bug tracking system for a complete list)

1) thorns CactusNet/http, external/TCPXX and external/RemoteIO
are not compiling under most platforms other than SGI or
Intel Linux

2) Cactus fails to compile for thorns with many grid variables
on the origin due to a problem with the C preprocessor.

3) There is a memory bug occuring occasionally in 1D ASCII IO

4) EllPETSc only compiles with MPI, but it doesn't tell you this

5) The testsuite checking technology isn't working under NT


==============================================================================


WHAT TO EXPECT IN BETA 4
========================

1) Fixes for all known bugs in Beta 3

2) A mechanism to get rid of all the warnings for unused
variables seen with Irix 6.5

3) Fuller implementation of 1D and 2D grid functions

4) Initial implementation of 1D, 2D and 3D arrays

5) Generic implementation of parameter choices for
elliptic slicings

6) Maximal slicing properly integrated into ADM


==============================================================================


TEST MACHINES NOTES
===================

* Intel Linux
o bifrost.aei-potsdam.mpg.de Redhat 6.0, Absoft
compiler, wo MPI
+ Remove: CactusPUGHIO/IOHDF5,
CactusElliptic/EllPETSc
+ Notes: CactusElliptic/EllPETSc compiled with
MPICH
o gullveig.aei-potsdam.mpg.de Redhat 6.0, Absoft
compiler, MPI LAM
+ Remove: CactusPUGHIO/IOHDF5,
CactusElliptic/EllPETSc
* Dec Alpha
o galar.aei-potsdam.mpg.de wo MPI
+ Remove: CactusElliptic/EllPETSc,
CactusPUGHIO/HDF5, external/RemoteIO,
CactusNet/http
* SGI 64 bit
o origin.aei-potsdam.mpg.de Irix 6.5, w/wo native
MPI
+ Remove: CactusElliptic/EllPETSc
+ Configure: HDF5=yes
HDF5_DIR=/usr/local/apps/hdf5
* SGI 32 bit
o nott.aei-potsdam.mpg.de Irix 6.5 wo MPI
+ Remove: CactusPUGHIO/IOHDF5,
CactusElliptic/EllPETSc
+ Notes: Problem with output for PsiKadelia
test
o null.wustl.edu Irix 6.2 wo MPI
+ Remove: CactusNet/http, CactusPUGHIO/IOHDF5,
CactusElliptic/EllPETSc, external/TCPXX,
external/RemoteIO
+ Notes: Gives a lot of warnings about data
alignment
* T3E
o pc.rzg.mpg.de w/wo native MPI
+ Remove:
CactusElliptic/EllPETSc,CactusPUGHIO/IOHDF5,
CactusNet/http, external/RemoteIO
+ Notes: dependencies went funny in IEEEIO,
problem with output for PsiKadelia test
* Windows NT
o saraband.aei-potsdam.mpg.de wo MPI
+ Remove: CactusNet/http, external/RemoteIO,
CactusElliptic/PETSc,
CactusPUGHIO/IsoSurfacer, external/TCPXX,
CactusWave/IDScalarWaveCXX,
CactusWave/IDWaveToyCXX
+ Notes: tested compilation only, problem with
perl testsuite runner
o ntsc-ts1.ncsa.uiuc.edu w/wo HPVM MPI
+ Remove: CactusNet/http, external/RemoteIO,
CactusElliptic/PETSc,
CactusPUGHIO/IsoSurfacer, external/TCPXX,
CactusWave/IDScalarWaveCXX,
CactusWave/IDWaveToyCXX
+ Configure: MPI=CUSTOM MPI_LIBS="mpi.lib fm.lib
advapi32.lib kernel32.lib wsock32.lib" MPI_INC_DIRS
="d:/apps/hpvm/include"
+ Notes: tested compilation only, problem with
perl testsuite runner
* HP
o exemplar.npaci.edu wo MPI, HP-UX 10.01
+ Remove: CactusNet/http, external/RemoteIO,
CactusPUGHIO/IsoSurfacer,
CactusElliptic/PETSc, external/TCPXX
+ Notes: dependencies went funny in TCPXX