cctbx multi-platform build 2004_09_22

Build tag: 2004_09_22_1424

Self-extracting binary distributions

cctbx+Python bundles are larger but completely self-contained.
cctbx only bundles use the Python that ships with the OS or require that a matching Python version is installed separately.
Use [stream] if your browser displays the [direct] download in a text window instead of saving it to a file.
Use cctbx.python to run scripts. The plain python command is not longer supported to avoid confusion with the Python version provided by some operating systems.

Platform	cctbx+Python	cctbx only
Mac OS 10.2.8 [Notes] Python 2.3	Python 2.3 must be installed separately	[stream][direct] 13.1 MB
Mac OS 10.3.9 [Notes]	Python 2.3 ships with OS	[stream][direct] 14.8 MB
Red Hat 8.0	Python 2.2.1 ships with OS	[stream][direct] 8.0 MB
Red Hat 8.0 Python 2.3.4	[stream][direct] 19.2 MB	[stream][direct] 8.0 MB
Red Hat 9.0	Python 2.2.2 ships with OS	[stream][direct] 8.0 MB
Red Hat 9.0 Python 2.3.4	[stream][direct] 18.9 MB	[stream][direct] 8.0 MB
Red Hat WS 3	Python 2.2.3 ships with OS	[stream][direct] 8.1 MB
Windows 2000/XP Python 2.2.3	A bundle with a more recent Python is available	[stream][direct] 5.6 MB
Windows 2000/XP Python 2.3.4	[stream][direct] 8.8 MB	[stream][direct] 5.6 MB

[Other platforms]

Download the self-extracting binary distribution for your platform to any new, empty directory. Under Unix selfx bundles are installed with the perl command (which is available by default on all supported platforms), for example:

    perl cctbx_redhat80.selfx

Under Windows exe distributions are installed simply by running them.

The installer must be run on the platform where it is used. The installed package may be used over the network, but only if the pathnames to the installation directory are identical on the server and the client.

Uninstalling a bundle is as easy as using rm -rf * in the installation directory under Unix or dragging the installation directory to the Recycle Bin under Windows. The Windows registry is not modified by the installation.

[Installation overview]

Self-extracting cctbx and Python 2.3.4 sources for Unix

cctbx_python_234_bundle.selfx [stream][direct]
21.5 MB

Download the file and run the following command in any new, empty directory:

    perl cctbx_python_234_bundle.selfx

This installs Python 2.3.4 and all cctbx modules from scratch.

The install script will prompt for the number of available CPUs. To avoid the prompt (e.g. for scripting the installation), add the number of available CPUs to the command line. E.g.:

    perl cctbx_python_234_bundle.selfx 4

Alternative for manual installation:
cctbx_python_234_bundle.tar.gz [stream][direct]
21.5 MB
[Installation instructions]

Self-extracting cctbx sources for Unix

cctbx_bundle.selfx [stream][direct]
13.3 MB

Download the file and run the following command in any new, empty directory:

    perl cctbx_bundle.selfx

This installs all cctbx modules from scratch. Python 2.2.1 or higher must be pre-installed on the target machine. Known to work under Redhat 8.0 or higher with the Python version that ships with the operating system (/usr/bin/python). The first python on PATH is used. To install with a different, specific python, add the full path to the command line, e.g.:

    perl cctbx_bundle.selfx /usr/local/bin/python

Alternative for manual installation:
cctbx_bundle.tar.gz [stream][direct]
13.3 MB
[Installation instructions]

Changes

global replacements:
  - 00000 -> False, 0001 -> True (Python 2.2 compatibility is achieved
    via the new libtbx.forward_compatibility module)

iotbx.pdb.parser:
  - new columns_73_76_evaluator
    - decides automatically if columns 73-80 can be ignored because
      they just contain the PDB ID and the record number

new iotbx.pdb.atom
new iotbx.pdb.interpretation
new mmtbx.geometry_minimization

new iotbx.wildcard:
  - tcsh-like wildcard matching (meta characters: *?[^])

iotbx.crystal_symmetry_from_any:
  - extract unit cell parameters from xplor maps

cctbx.geometry_restraints:
  - new bond_sorted_asu_proxies.show_sorted_by_residual()
  - new inverse_power_repulsion_function
  - repulsion_function renamed -> prolsq_repulsion_function
  - separate computation of nonbonded_sorted_asu_proxies and
    bond_sorted_asu_proxies to avoid excessive runtimes caused
    by very large bond.distance_model
  - automatic determination of nonbonded_distance_cutoff

cctbx.xray.structure:
  - new selection() method

cctbx.miller.array:
  - apply_selection() -> select()
  - new phase_integrator() for the conversion of Hendrickson-Lattman
    coefficients to centroid phases and figures of merit
  - expand_to_p1() now also works for Hendrickson-Lattman arrays
  - phase_transfer() now also works with Hendrickson-Lattman array
    as phase_source

cctbx.hendrickson_lattman:
  - new constructor given centroid phase and figure of merit
  - new operator+ and operator+=

cctbx.array_family.flex.hendrickson_lattman:
  - new constructor given centroid phases and figures of merit
  - support for phase combination (via __add__ and __iadd__)
cctbx.array_family.xray_scatterer:
  - new extract_u_cart(), set_u_cart()

cctbx.xray.scatterer:
  - bug fix: u_star eigenvalue filtering was not applied to
    sites in general positions

cctbx.crystal.pair_asu_table:
  - additional add_pair(pair) overload to facilitate easy combination
    with the cctbx.crystal.neighbors_fast_pair_generator.

cctbx.sgtbx.tr_vec, cctbx.sgtbx.rot_mx:
  - Python 2.4 compatibility: __float__ removed. Use as_double() instead.

scitbx.lbfgs:
  - new exception_handling_parameters

scitbx.math:
  - new signed_phase_error(), phase_error(), nearest_phase() functions,
    all in both scalar and vector versions

scitbx.array_family.flex:
  - new flex.bool(size, iselection) constructors
  - new flex.union, flex.intersection functions (returning flex.bool)
  - items() and indices() member functions removed (use enumerate()
    and xrange() instead)
  - new flex.extract_double_attributes()

scitbx.matrix:
  - Python 2.4 compatibility: __float__ -> as_float

scitbx.stl.map:
  - new stl_string_stl_vector_unsigned, int_stl_vector_unsigned

libtbx:
  - new libtbx.forward_compatibility injects False, True and enumerate
    into the __builtins__ namespace if not already defined

libtbx.refresh:
  - bug fix: use $LIBTBX0... only if defined; this works around
    problems caused by a blank PYTHONPATH under Mandrake Linux 10.0

libtbx.configure:
  - save build_options in libtbx_env and restore on warm_start

Latest Boost CVS (2004_09_20_1939)
Latest SCons CVS (2004_09_20_1712)