Limitations of the Fortran version of ATOMS

Development of ATOMS began many years ago as a Fortran program. Because of many limitations imposed by the choice of Fortran as the programming language, in 1997 I started rewriting ATOMS. The language I chose was Perl. While we could debate at great length the merits of choosing that language or some other language, the topic that deserves no discussion is that the Fortran version has many crippling limitations. Regardless, of my choice of languages, it was necessary for me to abandon Fortran so that I could address these many problems.

Most of the problem reports I get from users of the old Fortran version fall into one of the categories discussed below and are fully resolved in the new version.

Here is a partial list of the issues that are resolved in the current version of ATOMS that are unresolved (and never will be resolved by me) in the Fortran version:

• Very poor handling of monoclinic space groups. The Fortran version does not handle any but the standard setting. Making the Fortran version handle other monoclinic cell settings would have required altering the incredibly cryptic engine that interprets space group symbols or to perform the complex transformations between settings. Much better was to write a better interpreter for the space group symbols.

• Limited sizes of atom lists in the ATOMS and FEFF input files. Well ... Fortran has no dynamic memory allocation. The only solution for large atom lists was to edit source code and recompile. Perl has dynamic memory allocation. 'Nuff said.

• Perl is an excellent language for GUI and Web application programming. Fortran sucks for that. Atoms now exists in GUI and Web forms. Again, 'nuff said.

• The number of input file options in the the Fortran version which controlled small details of how the FEFF input files were written was expanding out of control. The new version used a template system for formatting output which is vastly more powerful, flexible, and extensible than piles of obscure input file options.

• Perl has object oriented capabilities. Consequently I was able to reuse much of the code from ATOMS to write programs to perform such interesting chores as powder pattern simulation, DAFS simulation, and calculations based on tables of absorption coefficients. All that would have been much more work in Fortran.