Release notes

Version June 30, 2024

• Major revision of the INIT step for increased execution speed.
• Modifications of the table GAIN.cbf. The gain values are now estimated at every 4th detector image pixel and multiplied by 1000 for storage as integer in the compressed file GAIN.cbf.
• The background values for each image pixel are multiplied by 1000 for storage as integer in the compressed files BKGINIT.cbf and BKGPIX.cbf, respectively.
• Input parameter STRONG_PIXEL= (previously used by the COLSPOT step) is now obsolete and has been replaced by the parameter SIGNAL_PIXEL= affecting steps INTEGRATE and now COLSPOT as well.
• Input parameters MINIMUM_FRACTION_OF_BACKGROUND_REGION= (previously used by the INIT step) is now obsolete.
• Input parameters DATA_RANGE_FIXED_SCALE_FACTOR= is now used only by INTEGRATE. In the INIT step it has no effect anymore.
• A new method for estimating the background in each image pixel has been implemented in the COLSPOT and INTEGRATE steps of XDS. Visual feedback of the computations is now provided by the control images SHOW_SPOT.cbf, SHOW_BKG.cbf, SHOW_HKL.cbf for a selected image specified by the new input parameter SHOW_IMAGE_NUMBER=
• Documentation of old reflection file formats DIRECT ANOMAL NORMAL OLDHKL UNIQUE that are still accepted by XDSCONV, XSCALE.
• A new independent utility program, GEOCORR is provided that computes correction tables used for analysis of electron diffraction patterns by XDS or nXDS.
• A new independent utility program, REZEL is provided that analyzes crystal symmetry as specified by given space group number and unit cell parameters.

• Error correction of the automatic determination of reflection shape in case of extremely sharp diffraction spots.

Version June 30, 2023

• Avoidance of strange characters in standard output upon reading images using the DECTRIS neggia library. (Problem reported by Aaron Finke).
• Maximum length of file names accepted by XDSCONV and XSCALE increased to 256 characters.
• Automatic removal of unused memory space in XSCALE.

Version January 10, 2022

• Correction of a bug in the IDXREF step that could lead to an odd indexing failure (error reported by Clemens Vonrhein).
• Values for the input parameters TEST_RESOLUTION_RANGE=, MIN_RFL_Rmeas=, and MAX_FAC_Rmeas= as used in the CORRECT step of XDS are shown in CORRECT.LP. Default values for these parameters are defined in the documentation.
• An additional version of the XDS package, XDS-Apple_M1, is provided for running natively under macOS Big Sur (or newer) on an Apple Silicon processor.

• Correction of a numerical problem in controlling the acceptance of reflections in the CORRECT step of XDS by the input parameter INCLUDE_RESOLUTION_RANGE=. (problem reported by Jan Wollenhaupt).

• Correction of a type mismatch in the routine linking the input of data images to an external library routine.
• Correction of reported indices for the difference vector clusters in the special 2-dimensional case in the IDXREF step.
• Exchange of 3 instructions to make the source code of the XDS package strictly obeying the Fortran 90 language definition.

• Correction of an error that might allocate processor threads to an empty set of images in the COLSPOT step.
• Optimized assignment of threads for a large multi-core processor system during the COLSPOT step.
• Change of the meaning of the correlation factors between data sets reported by XSCALE. Previously, these factors indicated the success of radiation damage corrections as correlation between mean intensities of data sets as function of resolution. The correlation factors now reported are calculated between intensities of equivalent reflections from each data set pair which is the more intuitive view (pointed out by Oliver Weiergräber and Kay Diederichs).

Version February 5, 2021

• Modification in the generic library routine used by XDS to avoid a frequent error message (Kay Diederichs & Sebastian Thorarensen).
• Construction of image file names was rendered thread-safe for the parallel version of XDS (Kay Diederichs).
• Prevention of a numerial overflow error in XSCALE (rarely occurring) by using double precision for a few variables.
• Additional information provided by XSCALE in the header of each scaled output file. For each input file it is reported whether reflections are merged and whether Friedel pairs are considered as different reflections or not.
• Correction of a reference in the XDS documentation describing the input parameter INDEX_MAGNITUDE= (Clemens Vonrhein).

• The input parameter EXCLUDE_DATA_RANGE= is provided now to the CORRECT step as well. This allows exclusion of reflections recorded by specific images without rerunning the INTEGRATE step.
• Correction of a bug in reading .h5 files by program 2cbf (Kay Diederichs).

Version January 31, 2020

• The routine for collecting results from processing batches of images during the INTEGRATE step was revised and enriched by the input parameter MAXIMUM_IMAGE_GAP_FOR_ADDING_PARTIALS=. This new parameter controls the addition of partial intensity contributions to the same reflection from adjacent image batches.
• Correction of code that could lead XDS to crash in the CORRECT step upon reading the reflection input file INTEGRATE.HKL (problem reported by James Holton and Kay Diederichs).

• Error correction in the file headers of INTEGRATE.HKL, XDS_ASCII.HKL. Here, the wave vector was given instead of the direction cosines for the parameter INCIDENT_BEAM_DIRECTION=. In case of short wavelengths, typical for electron diffraction experiments, this had lead XDS to crash upon reading the parameter in free format due to missing blanks separating the components (problem reported by Paul Klar).

Version March 15, 2019

• In the special case of a user supplied reading routine, specified by the input parameter LIB=, expansion of compressed images is now up to the external library routine.

• Changed default setting for the parameter MERGE_TREE=0.2 used in the IDXREF step. Too small values could lead to an augmented lattice basis that might prevent recognition of the correct unit cell and crystal symmetry (Problem reported by Karin Fritz-Wolf).
• The DEFPIX step now accepts a completely vanishing background (Problem reported by Tim Gruene).

• Bug correction in XSCALE. The problem caused a lower quality of the scaled data sets (reported by Kay Diederichs).

• Error message was added in case of undefined input parameter OSCILLATION_RANGE=.
• The default for the input parameter MAXIMUM_NUMBER_OF_PROCESSORS= MAXCPU, which defines the maximum number of !$omp-threads, is now provided by the hardware of the node executing each independent job of XDS. This allows XDS to better exploit the capabilities of an inhomogeneous processor cluster during the COLSPOT and INTEGRATE steps (modified by Kay Diederichs).
• A new input parameter SNRC=0.1 has been introduced to XDS and XSCALE which allows the user to control the minimum required I/Sigma for reflections to be used for the determination of scaling/correction factors.
• The meaning of the parameter MINIMUM_I/SIGMA= has changed. It now only selects sufficiently strong reflections for inclusion in the global refinemet in the CORRECT step of XDS. Its former additional role in selection of reflections for the determination of scaling/correction factors has been transferred to the separate parameter SNRC=.
• The new input parameter BATCHSIZE= is available to define the rotation range covered by each batch of images. This allows a better control of the number of correction factors used in XSCALE and the CORRECT step of XDS.
• The input parameter INCLUDE_RESOLUTION_RANGE= is now given priority in case this parameter is also present in the header of the reflection input file.

• Reduced execution time (~10%) for the INTEGRATE step of XDS by using the option for buffered I/O for compilation.
• Selection of a more robust low-intensity cut-off for accepting weak spots in the COLSPOT step. This mitigates the influence of "hot" detector pixels on indexing (reported by Clemens Vonrhein and colleagues at Global Phasing Ltd.)

Version January 26, 2018

• Revival of the refinement routine used by Version June 1, 2017 (BUILT=20170615). The old routine is more suitable to correct errors in the specification of the detector distance, often improving final data quality as well.
• Slight changes in the meaning of parameters controlling difference vector clusters in the IDXREF step; RGRID=, SEPMIN=, CLUSTER_RADIUS=.

• Improved algorithm in the IDXREF step for recognizing a lattice basis of unknown crystals. This replaces BUILT=20180302 which has been reported by Clemens Vonrhein and colleagues at Global Phasing Ltd. to collapse in some cases.

• The new input parameter MERGE_TREE=, effective in the IDXREF step for unknown crystals, allows the user to select translated reciprocal lattices for merge with the dominant one.
• XSCALE accepts the new input parameter PRINT_CORRELATIONS= which can be used to prevent printing long lists of pair correlation factors between the scaled data sets.

• Improved estimate for the default value for the raster step size in the 3D-histogram used for locating difference vector clusters in the IDXREF step (input parameter RGRID=).

• Improved error handling during the INTEGRATE step caused by the absence of strong spots for determination of reflection profile parameters.
• XSCALE accepts the new input parameter REIDX_ISET= which allows to reindex reflection indices specifically for each input data set.

Version Novenber 11, 2017

• XDS now accepts the new input parameter RGRID= allowing the user to define the grid size of the histogram for finding difference vector clusters in the IDXREF step - overriding automatic defaults.
• Code improvement to avoid occasional program collapse in the attempt to index a bogus set of spots.
• The file GXPARM.XDS will not be generated anymore in case the refinements in the CORRECT step were unsuccessful.
• The header of each output file generated by XSCALE now contains the cell constants of each included input data set. This helps detecting non-isomorphism and identifying compatible subsets of input data.
• Improved documentation for the input parameters of XSCALE.

• XDS now handles images obeying the definition of TIFF Revision 6.0

Version June 1, 2017

• XDS now accepts the new input parameter CLUSTER_NODES= for specification of processor nodes in a computer network
• Automatic assignment of specified processors of a computer network for minimizing the wall-clock time for running XDS. This implicates slight changes in the meaning of the input parameters MAXIMUM_NUMBER_OF_PROCESSORS=, MAXIMUM_NUMBER_OF_JOBS=, SECONDS=
• Work-around to prevent XDS from collapse in a cluster of multiprocessor nodes (reported by Michael.Fodje@lightsource.ca)
• Bug correction in the generic_reading routine for images with name extension .h5
• Corrected documentation for the input parameter REFINE(INTEGRATE)=
• Documentation of the input parameters RELRAD= and RELBET=.

• Revival of previous effects of the input parameters MAXIMUM_NUMBER_OF_PROCESSORS, MAXIMUM_NUMBER_OF_JOBS for scheduling computations by the user. This might simplify installation of the new XDS version in a computer network with no explicit naming of its CLUSTER_NODES.

• Change of the refinement protocol and of defaults for the refined parameters at various stages of processing to reduce the risk of divergence of the procedure.
• Revival of previous behaviour of XDS allowing parallel processing in case data processing overlaps with the measurements.
• The shell script forkxds also provides the parameter value maxcpu (again) to simplify installation of XDS on a supercomputer cluster using a SLURM workload manager.
• Prompt error message upon calling XDS when the specified dynamic library file does not exist.
• Correction of the consequences of a rounding error that could cause XDS to collapse in the INTEGRATE step.
• Clarification of the documentation on the parameter NAME_TEMPLATE_OF_DATA_FRAMES= in case a dynamically linked library is used for accessing images.

• Bug correction in the COLSPOT step that sometimes refused too many weak spots.
• The input parameter CLUSTER_NODES can now accept 2048 characters for naming the nodes of the cluster.
• Modification of the shell script forkxds which now uses the full path names of the invoked executables.
• Negative values for the input parameter MINIMUM_VALID_PIXEL_VALUE are forbidden now.

• Addition of an error message in case of incomplete parameter specification of multi-segment detectors. For these instruments, this now prevents XDS from collapse at later stages of processing.

Version November 1, 2016

• XDS, MERGE2CBF, and 2CBF now accept the new input parameter LIB= that specifies the name of an external library for reading images. This new feature renders the XDS package independent from technical details of future detector image formats.
• New input parameter GAIN= allows to fix the conversion factor between image pixel contents and equivalent counts (for a pixel detector you could set GAIN=1.0).
• Improvements in speed of XDS by parallel execution of the INIT step.
• Increase from 32 to 99 for the maximum number of cpu's that can be employed by the parallel version of XDS (MAXIMUM_NUMBER_OF_PROCESSORS=).
• Improved background determination for estimating reflection intensities.

• Bug fix in XDS in representing large background values in BKGINIT.cbf.
• Code changes in the INIT step of XDS to prevent a possible stack overflow and subsequent collapse of the program.
• Fine tuning in determination of low background for estimating reflection intensities in the INTEGRATE step of XDS.
• Bug correction in XDSCONV when using the output format keyword CCP4_I+F for saving structure factor amplitudes as well as intensities in MTZ files.

• Bug correction in XDSCONV in generating output files suitable for CCP4. The bug could lead sometimes to reflection records IMEAN,SIGIMEAN=0,-1
• Generalization of XDS for naming image templates accessed by an external library routine specified by LIB=

Version May 1, 2016

• XDS now accepts a new input parameter EXCLUDE_DATA_RANGE for excluding data images from processing.
• The new input parameter FIT_B-FACTOR_TO_REFERENCE_DATA_SET forces the isotropic temperature factor of the processed data to match the corresponding one of a reference.
• Error correction in calculating the default value for the XDS parameter NUMBER_OF_IMAGES_IN_CACHE=. The error could lower the efficiency of caching images in some cases.
• XDSCONV now has the option to save structure factor amplitudes as well as intensities in MTZ files using the new output format keyword CCP4_I+F in specifying the parameter OUTPUT_FILE=.

• fixes occasional problems in IDXREF by recompilation of the package using the -fpe0 flag of the INTEL ifort compiler.

• improves efficiency of the image cache
• corrects a bug in XDSCONV for converting XDS files to filetype CCP4 when using GENERATE_FRACTION_OF_TEST_REFLECTIONS. This special combination leads to a loss of half the data. This bug is absent from all other cases and all versions of XDSCONV released before May 1, 2016.

Version October 15, 2015

• Improvements in speed of XDS by implementing a cache to access detector images more efficiently (suggested by Kay Diederichs). A new input parameter, NUMBER_OF_IMAGES_IN_CACHE=, is provided in XDS.INP for explicit control of the cache size by the user.
• Simplification in the method for calculating the initial value of the mosaicity. The new method avoids a numerical problem that has lead to a crash during the INTEGRATE step of XDS in rare cases.
• The value of the input parameter OSCILLATION_RANGE= has been restricted to be a positive multiple of 0.0001 degrees.
• Error correction in XSCALE (problem reported by Alexey.Rozov@igbmc.fr). When merging datasets numbering several millions of reflexes XSCALE collapsed (floating overflow) in the calculation of dataset statistics.

Version June 17, 2015

• Improvements of XDS in the determination of the integration region of diffraction spots recorded at large detector distances (above 3500 mm) and short wavelengths (below 0.02 Å).
• Patch interpreting pixels of EIGER detector images with suspiciously high values as broken.
• Format changes at various places in the programs that allow for sufficient space for printing large numbers.
• Improvements in XSCALE by listing incompatible data sets or those lacking overlap with the main set (suggested by Kay Diederichs).

Version March 1, 2015

• Improvements of XDS for low dose data.
• Improvements of XSCALE for efficient scaling of a large number of data sets. To offer a possibility to suppress the generation of numerous control images (which would hardly ever been looked at) the new input parameter SAVE_CORRECTION_IMAGES=, was introduced.
• A simple user interface was introduced for controlling the refinement of segment orientation and translation for a multi-segment detector; it uses the two new input parameters DEFAULT_REFINE_SEGMENT, and MINIMUM_NUMBER_OF_REFLECTIONS/SEGMENT.
• The former (misleading) keyword DISTANCE has been replaced by the keyword POSITION for specifying refinable parameters during the IDXREF, INTEGRATE, and CORRECT steps (REFINE(IDXREF)=, REFINE(INTEGRATE)=, REFINE(CORRECT)=, DEFAULT_REFINE_SEGMENT=, REFINE_SEGMENT=).
• A utility program PIX2LAB was added as a stand-alone program that helps finding the laboratory coordinates for a pixel specified by the user according to a given input file XDS.INP. This utility is thought to help the user verifying the correctness of XDS.INP for complicated multi-segment detectors.
• Error correction in the refinement routines for multi-segment detectors (problem reported by Claus Flensburg, Clemens Vonrhein, and Gerard Bricogne).

Version November 3, 2014

• XDS accepts XZ-compressed image files recognized by the .xz name extension
• XDS accepts images obtained by the EIGER (DECTRIS) detector. The conversion from HDF5-file format to .cbf requires installation of the separate program H5ToXds available from DECTRIS ( http://www.dectris.com).
• Correction of a bug in XSCALE that prevented scaling data sets containing very few reflections. The program terminated with the incorrect message !!! ERROR !!! INSUFFICIENT NUMBER OF COMMON STRONG REFLECTIONS (error reported by Dmitry Popov)
• Correction of an error in calculating items ALF0,BET0,ALF1,BET1 of each reflection record in file INTEGRATE.HKL (error reported by Claus Flensburg & Clemens Vonrhein)
• Correction of a 32 bit integer overflow error in the INTEGRATE step that let XDS silently terminate (error reported by Claus Flensburg). This error occurred when trying to allocate a large number of finely rastered reflection profiles in the learning phase.

Version January 10, 2014

• Adaptation to read Rigaku's modified image format (SMV) for 200/300K pixel detectors
• Correction of a bug in the determination of reference profiles in the INTEGRATE step of XDS that could cause a program crash.
• Correction of a bug in background determination in the INIT step that could result in the (wrong) error message 'SPECIFIED DATA IMAGES UNAVAILABLE'
• Increase from 8 to 9 positions for printing Rmeas in CORRECT.LP and XSCALE.LP. This simplifies parsing the output by other programs.
• Input file names up to 120 characters are now accepted by XSCALE

Version November 11, 2013

• A possibility for omitting specified reflections from file INTEGRATE.HKL in the CORRECT step has been included. These reflections are collected in the new file FILTER.HKL Typically, the file is generated by a dedicated user program to handle special experimental situations in which parts of the detector become obscured.
• Bug correction in the XYCORR step of XDS in the case of a multi-segment detector at vanishing DETECTOR_DISTANCE. (Bug reported by Claus Flensburg.)
• Correction of a bug in IDXREF that sometimes caused a failure to recognize a given unit cell basis due to insufficient sampling of possible basis orientations.

Version March 30, 2013

• Extensive modifications of XDS for processing data from a generalized detector consisting of any number of rectangular segments in some fixed arrangement. For description of multi-segment detectors see section Coordinate Systems. 7 new parameters are introduced SEGMENT=, REFINE_SEGMENT=, DIRECTION_OF_SEGMENT_X-AXIS=, DIRECTION_OF_SEGMENT_Y-AXIS=, SEGMENT_ORGX=, SEGMENT_ORGY=, SEGMENT_DISTANCE=.
  The new parameters are irrelevant for single-segment detectors.
• Parameters INCLUDE_RESOLUTION_RANGE=,EXCLUDE_RESOLUTION_RANGE= are now activated for filtering spots used in the IDXREF step.
• Bug correction in a lattice recognition routine that could end up in a non-terminating loop (reported by Claus Flensburg).
• default value corrected to WILSON_STATISTICS=TRUE

Version September 26, 2012

• Bug correction in the IDXREF and CORRECT step of XDS. In some cases basis vectors specified by the user in XDS.INP were ignored, using non-equivalent alternative settings instead. The bug was introduced in the previous version July 4, 2012 while earlier versions were o.k. (bug reported by Clemens Vonrhein).

Version July 4, 2012

• The four user input parameters REFLECTING_RANGE=, REFLECTING_RANGE_E.S.D.=, BEAM_DIVERGENCE=, and BEAM_DIVERGENCE_E.S.D.= can be independently fixed or automatically determined by XDS.
• The new parameter, DATA_RANGE_FIXED_SCALE_FACTOR= i1 i2 s, was introduced to allow an explicit specification of a scale factor 's' for each image in the data range 'i1' to 'i2'. Any number of such parameters can be specified. This new parameter replaces the previous one FIXED_SCALE_FACTOR= TRUE
• Removal of outliers from the comparison of the new data with a reference data set in the CORRECT step. The outliers have sometimes caused low correlation factors being reported.

Version March 15, 2012

• Replacement of Rmrgd-F by a correlation factor CC(1/2) between random half-datasets for reporting results in CORRECT and XSCALE. CC(1/2) is a useful indicator of crystallographic information in each resolution shell. (Karplus & Diederichs).
• Correction of a bug in the INTEGRATE step that lead to small inaccuracies in the percentage of observed reflection intensities (item PEAK in INTEGRATE.HKL).

Version March 5, 2012

• Correction of a bug in the INTEGRATE step that sometimes caused a reading error (reported by Michel Fodje, Canadian Light Source).
• Correction of a bug in the INTEGRATE step that lead to inaccurate spot boundaries for rotation images with a large oscillation range. (pinpointed by Kay Diederichs).

Version January 19, 2012

• The input parameter UNTRUSTED_QUADRILATERAL= has been introduced to simplify exclusion of shaded pixel regions that cannot be enclosed by border lines parallel to the detector's X and Y directions.
• The input parameter MINIMUM_FRACTION_OF_BACKGROUND_REGION= controls the minimum allowed size of background islands on the detector.

Version December 31, 2011

• Improvements in the INTEGRATE step by modelling of individual reflection routes through the Ewald sphere (mainly affecting spots close to the rotation axis).
• Improvements in the success rate of the IDXREF step.
• Routines for processing BRUKER SMART6000 images have been reactivated. (Note, that XDS cannot correct these images for spatial distortions!)
• The input parameter FIXED_SCALE_FACTOR= has been introduced to handle data images with extremely low background. Nearly vanishing background such as obtained from extremely fined-sliced data from the PILATUS detector could lead to underestimated errors of the integrated intensities. The optimal oscillation range need not be less than half the crystal's mosaicity.
• The input parameter WFAC1= is now provided also to XSCALE allowing the user additional control on accepting symmetry related reflections for determination of correction factors.

Version December 6, 2010

• The new input parameter SECONDS= was introduced to check for the existence of a data image in units of seconds rather than minutes. This new parameter replaces the old parameter MINUTE= which is no longer accepted.
• For stopping a running xds job more gracefully than by issuing CTRL-C the user could now create a file named 'stop' in the XDS processing directory by the command 'touch stop'. Each of the parallel threads then refuses to process any more data images.
• Correction of an error that prevented the reading of CBF-type files that are compressed by gzip, bzip2, compress.
• IDXREF now works more robustly with few reflections, e.g. when indexing a single image.
• Improvements in decision making for resolving ambiguities in the unit cell basis vectors if a reference data set is present. (Effects CORRECT and XSCALE)
• The recently developed fast scaling method (Kabsch, 2010b) was extended and is now again using the smooth scaling equations (Kabsch, 1988b), generally improving data quality.

Version May 10, 2010

• Reduction of fluctuations in scaling factors of adjacent images with a very low background (contributed by Kay Diederichs).
• The new input parameter UNTRUSTED_ELLIPSE= was introduced to simplify exclusion of detector pixels behind the beam stop.
• The highest signal-to-noise value that can be obtained by the experimental set-up is reported in the CORRECT step as indicator of instrument error.

Version February 3, 2010

• Correction of convergence problems in the refinements of crystal parameters.
• Three parameters were created: MINIMUM_FRACTION_OF_INDEXED_SPOTS=, MAXIMUM_ERROR_OF_SPINDLE_POSITION=, REJECT_ALIEN=. The first two control automatic decision making for success or failure of the IDXREF step. The last one controls automatic removal of "alien" reflections from the final data.

Version December 28, 2009

• Error correction in determination of very low background for the PILATUS pixel detector.

Version January 30, 2009

• Automatic recognition of image file formats for most detectors currently in use.
• XDS now uses 32 bit integers throughout to code image pixel values. Control images generated by XDS are efficiently saved as 32 bit-pixels in the CBF format, a byte-offset variant of the CBFlib format. Such images are indicated by the file name extension ".cbf".
• The graphics program VIEW is no longer distributed with the XDS package. It is recommended to use the open source program XDS-Viewer that can display CBF-formatted images. Other file formats are supported through the external conversion program 2cbf that comes with the XDS package. XDS-Viewer is distributed under the GNU General Public License and must be downloaded separately.
• Executables of the XDS package for SGI (IRIX), Compaq alpha (Tru64), and Apple Power Mac G4/G5 (OSX) are no longer provided.

Version September 10, 2008

• Error correction in the CORRECT step of XDS. The error could prevent the automatic selection of the correct unit cell setting with respect to that of an external reference data set.

Version July 21, 2008

• XDS now accepts images from the Rigaku Saturn 944+ CCD detector.

Version June 2, 2008

• A subroutine to support space group determination was added to the CORRECT step. Automatic assignments are based on a few decision constants that can be modified by the user in XDS.INP. These new input parameters are called MAX_CELL_AXIS_ERROR=, MAX_CELL_ANGLE_ERROR=, TEST_RESOLUTION_RANGE=, MIN_RFL_Rmeas=, MAX_FAC_Rmeas=.
• Bug correction in XDS. Error in calculating reflection boundaries for centered lattices that could lead to inferior results at high resolution in the INTEGRATE step (pointed out by Kay Diederichs).

Version December 6, 2007

• Three new input parameters for use in the IDXREF step, UNIT_CELL_A-AXIS=, UNIT_CELL_B-AXIS=, UNIT_CELL_C-AXIS= have been introduced to help the user maintaining a consistent indexing scheme between several data sets taken from the same crystal in the same orientation.
• The weighting scheme for obtaining reference profiles in the INTEGRATE step that was introduced in the previous version proved inadequate for data from the PILATUS pixel detector and has now been replaced by the old weighting scheme.
• Bug correction in XDS. The input parameter DATA_RANGE= was ignored in the CORRECT step.
• Bug correction in XSCALE. The column OBSERVED reported a too small number of observed reflections in each resolution shell.

Version November 10, 2007

• Minor change in weighting 3D reflection profiles in the determination of references used for profile fitting in the INTEGRATE step. This may result in a slight improvement of the final intensity data.
• Small modification in estimating errors of reflection intensities resulting in a reduction of random fluctuations of the errors.

Version October 10, 2007

• Changes in the INTEGRATE step of XDS to reduce memory requirements. This solves a problem in processing fine-sliced data images.

Version August 15, 2007

• Minor bug corrections and simplifications in the use of XDSCONV and XSCALE. Some of the input parameters, such as space group and cell constants, are now extracted from the header of the input file by default. Note that the definitions of the parameters INPUT_FILE= and OUTPUT_FILE= have changed and a new parameter INCLUDE_RESOLUTION_RANGE= has been included in XDSCONV and XSCALE. Thus, files XSCALE.INP and XDSCONV.INP that worked for the previous versions of the programs are likely to need a modification.

Version July 5, 2007

• Bug correction in CORRECT. The intervals for reporting the statistics MEAN INTENSITY AS FUNCTION OF SPINDLE POSITION WITHIN DATA IMAGE were incorrectly chosen.
• Bug corrections in XDSCONV that affected the conversion to CCP4 format. XDSCONV has been extensively revised and now uses the keywords CCP4_F and CCP4_I instead of the old specifier FALL and IALL, respectively.

Version June 12, 2007

• Improvements in CORRECT and XSCALE in calculations of correction factors and error model parameters of reflection intensities.

Version May 8, 2007

• The default value of the input parameter STRICT_ABSORPTION_CORRECTION= has been changed from TRUE to FALSE.
• Minor changes in CORRECT and XSCALE in calculations of correction factors and error model parameters of reflection intensities.

Version March 15, 2007

• XDS now accepts images in the CBFlib format (new format keyword is CBF).
• The new input parameters SENSOR_THICKNESS= SILICON= were introduced to XDS to specify the finite thickness of the silicon sensor of the PILATUS pixel detector. They are used in the XYCORR and CORRECT steps to account for parallax- and sensitivity effects.
• The old input parameter AIR= has become optional as the appropriate, wavelength-dependent value will be calculated by XDS.
• The old input parameter MINPK= is effective now in the INTEGRATE as well as in the CORRECT step.
• Restrictions on the total number of parameters of types UNTRUSTED_RECTANGLE=, SPOT_RANGE=, and EXCLUDE_RESOLUTION_RANGE= have been removed.
• A new input parameter CORRECTIONS= has been introduced to allow the user to specify which corrections should be made in CORRECT/XSCALE

Version December 6, 2006

• New routines in CORRECT and XSCALE for rapid calculation of data correction factors. The correction factors are provided as color coded 2-D images that can be inspected with the VIEW program of the XDS-package.
• Improved diagostics in CORRECT and XSCALE for detecting anomalous signals in the collected data.
• The new input parameter MINIMUM_I/SIGMA= was introduced to XDS and XSCALE which defines the minimum signal-to-noise for reflections included in correction/scaling factor determinations.
• The new input parameter REFLECTIONS/CORRECTION_FACTOR= was introduced to XDS and XSCALE which defines the minimum number of reflections needed for the determination of each correction factor.
• The new input parameter MINIMUM_ZETA= allows the user to specify the size of the 'blind region' for excluding reflections close to the rotation axis.

Version August 18, 2006

• Correction of an old bug in IDXREF. In rare cases this bug could prevent the interpretation of the observed reduced cell vectors in terms of the conventional cell leading to failure of IDXREF.
• Correction of a bug in CORRECT introduced in the August 8 release. This bug could lead to a core dump if an external reference data set was used.
• First release of XDS for Apple MacBookPro (Intel Core Duo) on August 21, 2006.

Version August 8, 2006

• The correction model adopted in CORRECT and XSCALE in the June 2006 release turned out to be less robust than that of the older versions. For this reason the old model was revived.

Version June 2006

• The dynamic range of image pixel data that can be handled by XDS has been increased to 20 bits (previously 18).
• Faster execution of CORRECT and XSCALE, resulting from a new central data structure more suitable for parallel execution (Openmp).
• The CORRECT step has been modified to accept the additional input parameter NBATCH=
  which controls the number of absorption correction factors. Consequently, the action of the input parameter DELPHI=
  is now confined to the INTEGRATE step and no longer used by CORRECT.
• Bug fix in the reading routine of images from the saturn92 and RAXIS detector.
• Removal of restrictions for the total number of input files or reflections that can be processed by XSCALE.

Version November 2005

• Bug fix for the ADSC Q315 detector when the image header exceeds 512 bytes.

Version August 2005

• XDS now accepts images in the STOE format (Stoe & Cie GmbH, Darmstadt, Germany). To correct the images for spatial distortions a new parameter was introduced, STOE_CALIBRATION_PARAMETERS=.
  The routine for reading images in the STOE format as implemented in XDS is incomplete because pixel values above 65535 are always treated as overflows.
• The INTEGRATE and CORRECT steps of XDS as well as XSCALE now provide Schwarzenbach and Flack's psi-angle for each reflection which completely specifies diffraction geometry with respect to the unit cell axes.
• Correlation factors between observed and expected reflection profiles reported near the end of INTEGRATE.LP have been incorrect and came out too low (bug reported by Fred Vellieux, CNRS, Grenoble). The bug has no consequences on the final data set.
• The polarization correction factor was incorrectly calculated in the INTEGRATE step. The effect on the final data set is likely to be marginal in most situations. In case of doubt the INTEGRATE and CORRECT steps should be repeated with the new version of XDS.

Version May 2005

• The DEFPIX step has been modified to accept the input parameter EXCLUDE_RESOLUTION_RANGE=
  that was used previously only by the CORRECT step. The new feature allows to exclude ice-rings from the trusted region of the detector surface prior to the integration process.
• The INTEGRATE step has been substantially revised. Compared with the previous versions it requires only a moderate virtual address space even for very large data sets.
• The COLSPOT and INTEGRATE steps of XDS have been modified to make efficient use of Unix/Linux-clusters to reduce the wall-clock time needed for data processing. It is no longer required that the processors at each cluster node share the same address space. In case they do, XDS can exploit additional data parallelism at a more fine-grained level. The user can set (in XDS.INP) the number of nodes in the cluster by the new input parameter MAXIMUM_NUMBER_OF_JOBS=
  
• Obsolete input parameters (causing XDS to stop with an error message):
  RMAX=
  PROFILE_RANGE=