chem cryst
CRYSTALS Software
University of Oxford
-Crystals
+ Research
Contact
Lectures
+ Links
+ Gallery
CSD
Chemistry Dept
University of Oxford
chem cryst news
Many small changes in Regularise Replace thanks to a CRYSTALS workshop in Toulouse organised by Carine Duhayon and Laure Vendier. Cif generator now properly includes esds on refined hydrogen bonds, has incrreased figure-fields for cell parameters, and appends constraint and restraint information as _iucr_ data items at end of cif. Improved scaling of difference Pattersons - see forthcoming paper by Flack, Sadki, Thompson etc. Absolute configuration routine (based on PLATON with Ton's permission) extended to provide more diagnostics and additional data to be re-input as absolute-structure strengthening restraints (see paper by Howard etc above). Error tracking for "USE" files has been improved.
Collaboration with Ernesto Mesto (Dipartimento Geomineralogico - Università degli Studi di Bari) uncovered over-sights in the original 1979 code, which lost contributions when a twin component fell onto a centring absence. Difference maps with twinned data now seem to reveal hydrogen atoms very reliably.
Bob Carruthers worked with John Rollett and Keith Prout to produce the core data management and crystallographic subroutines upon which CRYSTALS is built. It was his vision and meticulous attention to detail which provided a kernel which still works well after 40 years of change in computing technology.
The Crystals code (over 1/3 million lines of FORTRAN, 26,000 lines of C++, and 28,000 lines of SCRIPTS) has been seriously overhauled to add the new Windows interface. Inevitably problems will crop up in rarely used routines and in situations that we haven't yet tested. By enrolling the help of users to help us to reproduce and fix these problems, we can rapidly make the program more robust for the rest of the community.
The new Windows version of CRYSTALS can be particularly difficult to debug because if there is an unhandled error in the underlying FORTRAN code which causes a crash, the whole program disappears from the screen, leaving no trace of what went wrong.
Fortunately, CRYSTALS writes several log and listing files while it is running and these can usually be used to identify the problem.
The log file
If you were typing in commands, or using commands from a file with a #use instruction, or runing a script, then the commands which preceeded the crash will be found in bfile.log in the working directory. Many problems can be caused by badly written scripts, often by INSERTing or STOREing too much text in the SEND command buffer. We will solve this shortly by adding extra checking to the SCRIPT processor code.
The monitor file
This file shows a brief summary of progress, it is virtually equivalent to
the text appearing on the screen.
The first file is named bfile#00.mon, the next
bfile#01.mon, bfile#02.mon, up to bfile#99.mon.
The listing file
This file starts as bfile#00.lis and follows the same numbering scheme as the monitor file. It provides a much more detailed output than that sent to the screen, including details of files being opened and closed.
The gui error log
Named script.log, this file records any errors in commands sent to the GUI.
If none of these files helps to narrow down the problem, and if it is reproducible, then send in a bug report form with a description of the problem.
Thanks for your help.
[Problem finding]
