Table Editor And Planner, Or: Teapot!

teapot is a spread sheet program for UNIX, which uses modern concepts besides being portable and extensible, for instance three dimensional tables and iterative expressions.

The most important feature is the funktional addressing of cells, unlike the traditional approach like A0, C5, etc. which is probably inherited from VisiCalc.

The basic features, which make teapot different from other spread sheet programs, are:

Instead of the unflexible and complicated syntax used to address the values of cells in traditional spread sheets, teapot uses for a more functional approach. This results in an easily understandable syntax and despite the very few basic functions, very powerful semantics.
In traditional flat spread sheets, you have to have the ability of hiding columns or rows to hide intermediate results. This indicates that two dimensions are not enough, which is why teapot is three-dimensional. The third dimension can be imagined like multiple layers on top of each other.
Unlike many other spread sheets, all values have a type, and all operations and functions check types. As a conclusion, adding a decimal number and a string will result in an error, instead of an (probably unwanted) result.
teapot offers very comfortable iterative expressions, modelled after the clocked expressions in the current PhD work of Jörg Wittenberger <joerg.wittenberger@inf.tu-dresden.de> at the University of Technology in Dresden, Germany.
The spread sheet file format is defined using XDR, which makes sure it is portable accross machines and there are the least possible number conversion errors, usually none. Additionally, a portable ASCII file format is supported to ease automatic generation of sheets by shell scripts; as well as importing sheets in CSV, SC and WK1 format.
teapot can easily be extended by new functions in C. Instead of just adding all functions offered by other spread sheets, as few as possible were implemented as canonical as possible. So, if a function is lacking, it means that so far nobody needed it or there is a more general way to solve the problem. Don't hesitate to send me email to solve such issues.
I use X/OPEN message catalog support for all messages, so producing a translated version is no big deal. Further localisation is planned. teapot comes with English, Dutch and German message catalogs.
Where possible, the UNIX philosophy of keeping things small and beautiful if being followed. That means, instead of PostScript/dvi/whatever output, there are interfaces to troff/tbl and LaTeX and HTML and CSV files can be generated. The latter is a very popular interface for graphics software.

To illustrate the above points, a few examples will be presented. One of the most important functions in teapot is @(x,y,z). Called with integer values, it evaluates to the value of the cell at the given coordinates. You can omit any of these values, which means the coordinates of the current expression will be used. Due to the functional approach, of course any integer value may be a literal or an expression which evaluates to a value of the type integer. Obviously, indexed accesses come for free.

Alternatively, @() can be called with a pointer to a cell as argument. Elements of this pointer data type refer to the position of a cell and are defined by labels or functions. Labels make expressions much more readable: Instead of A0+A1 (traditional) or @(0,0)+@(0,1), you can use @(Fix_costs)+@(Variable_costs) or, if you label columns and rows, @(x(Costs),y(Paper))+@(x(Costs),y(Tapes)). (x() and y() evaluate to the column and row of a pointer value.)

Another important element is eval(). This function evaluates to the value of the expression pointed to by the argument, but evaluation takes place in the context of the current expression. Instead of using many cells with the same expression, one cell could be labelled and the other cells only refer to its expression. If the expression has to be modified, only one modification has to be made.

teapot has no builtin programming language, because most programs written in such languages are hard to maintain. If you want a full programming language, you know where to find it. However, this does not mean that e.g. conditional evaluation is lacking. Conditional evaluation is possible through eval() with a pointer, where one coordinate contains one or more relational expression of which each evaluates to 0 or 1.

The mentioned clocked (iterative) expressions consist of two parts, called the reset and the clocked part. The very first time a cell is evaluated, the reset part will be used, and subsequent evaluations will use the clocked part. The reset expression 1 and the clocked expression @()+1 results in a cell, which will count upwards each time the sheet is clocked. It is quite easy to model the game of life or iterative computation of square roots as a spread sheet this way.

teapot is written in ANSI C and it should compile on any mostly POSIX compliant system, with the following two restrictions: SYSV curses is used for the user interface, so there is quick screen updating and function key support. The free SYSV curses clone ncurses works fine for that. Further, you need the Sun XDR library, but its code is also free available. teapot is known to work on Solaris 1.1.2, Solaris 2.3, HP-UX 9, Linux, FreeBSD 2.x and Ultrix 4.3. Porting to other systems is usually only a matter of changing the Makefile.

teapot is available via ftp as GNU zipped tar file with sources and documentation.

Michael Haardt