Samples of My Work

Ada

I was a long-time member of ACM's AdaTEC and then SIGAda, but I have yet to work on an Ada project. The only access I've had to an Ada compiler (aside from a one-week course I took in 1987) was an early Meridian compiler - running under MS-DOS on an XT clone! The compiler was surprisingly quick and it came with a good library. I wrote several simple utility programs, nothing fancy, from which the samples below are drawn.

alex.ada

A package of miscellaneous utility functions.

bopt.ada

A command line option processing package.

more.ada

A more(1) program in Ada.

ALGOL 60

Thanks to a book by Daniel McCracken, ALGOL was actually one of the first languages of which I had a reading knowledge. I later used it in the file structures course and the assembly language course at the University of Maryland; everyone else used FORTRAN. Norwegian University ALGOL (NU-ALGOL) was the compiler. The manual had a list of compile-time error codes and descriptions, including the following:

ERROR NUMBER	MESSAGE	POSSIBLE PROBLEM
22	Too many errors - compilation suppressed	Have you read the programmer's guide?
27	Internal error	The user has totally confused the compiler. Correct all other errors and try again.
40	*Warning* Do you want to compare constants?	Possible punching error.

C++

(Written in 1995, although some of the complaints are still valid.) I'm disillusioned with C++. In spring of 1995, a small team was formed to develop a telemetry data generator in C++. With great fervor, I jumped into developing C++ versions of various packages in my C General Purpose Library. Some of the insights I had while working on the C++ library I decided to retrofit in the C library; the networking packages, in particular, were redone. During the summer, all but one of us on the team were pulled off to work on other tasks.

Why was I disillusioned?

• Templates - were awkward to structure; the source text for the class had to be included directly or indirectly in the class header file. At the time, G++ supported templates in this manner; in the Multi development system some of the team were using, only the command-line compiler supported templates. Ada and Eiffel have long had generic functions, packages, and classes; what's taking C++ so long? (Admittedly, Ada and Eiffel also have a more sophisticated build environment than simply the standard UNIX linker.)

• Standardization - is a well-known failing of C++. At the time, G++ supported most, but not all, of the standard C++ header files and library routines; Multi wasn't even in the ballpark. Which compiler do I write for?

• Exceptions - were not well supported. C++ exceptions are more ambitious to be sure, but Ada and Eiffel have long had exception handling.

• Complexity - is C++. C sometimes needed a language lawyer; C++ needs a whole team of lawyers. You can't simply take the approach of some that only the "good" programmers should program in C++. We good programmers can't carry the weight of the programming world on our shoulders! (smiley :)

• Documentation - is necessary, but I never found or developed a standard for C++ source code documentation with which I was comfortable. The increasing role played by the header file in C++ results in a division of documentation between the header file and the source file.

VeryLong.cc

VeryLong.hh

A C++ implementation of multi-precision longwords.

COBOL

In the database course at the Univerity of Maryland, we wrote COBOL programs to access the Univac DMS 1100 CODASYL database. It was a great course (taught by Dr. Michael Brodie) and COBOL wasn't bad. Alas, I have no samples of my work - it was a long time ago!

FORTRAN-77

crsredit.for

Not my latest and greatest FORTRAN code, but this function made it easy to write full-screen, VT100 "GUI"s for a number of applications I wrote.

Pascal

Pascal? I could take it or leave it. Most of my upper-level computer science courses used Pascal; e.g., the data structures course, the compiler course, etc. I remember from the compiler course that string handling was a chore. The operating systems course used PDP-11 assembly language, but I would first write the programs in a Pascal-like PDL and then translate them to assembly language.

PL/M-86

A readable C!

tirjob.plm

A central task that handled I/O from multiple sources.

help.txt

Commands that the TIRJOB task understood.

8086 Assembly Language

Bleecchh!

memtst.a86

Memory diagnostics code.

msurst.a86

Power-up/Reset firmware.