If you’ve compiled the OpenSees source code, you may have noticed the
SRC/string directory, whose two files,
G3string.cpp, implement character strings as first-class objects, complete with overloaded operators and deep copies.
G3string is based on the String class, which I used for programming projects in my undergraduate computer science courses. In the
G3string.h header file you’ll see
#define _PLSTRING_H, which I’m fairly certain is a reference to Jo Ellen Perry and Hal Levin, authors of the C++ book we used in CSC 114 and 210 at Pine State University.
In graduate school, I thought the String class would be a good addition to G3–I mean, we should return String objects instead of
const char * and use
== instead of
strcmp, right? But, beyond recorders and parameters, core string functionality is only marginally useful for finite element implementation. This functionality becomes even less useful considering all the string libraries available in Tcl, and now Python.
I was in charge of the Visual Studio workspace at the time, so despite the limited utility of string objects in G3, I created a dedicated project for G3string alongside element, material, and the two dozen or so other projects. The string project is still in the OpenSees Visual Studio workspace today.
However, there’s never been a Makefile in the
SRC/string directory, making G3string a member of the OpenSees retirement home–the source code that remains in the repository but is either not compiled or compiled but not linked with the executable or the Python module. There’s more residents than you think.
To address this longstanding oversight, I added a Makefile to the
SRC/string directory via pull request #185. The G3string class compiles just fine on Windows and Linux and is now one step closer to coming out of retirement. For what reason, I don’t know.