Unci: Unit testing with a clean interface

I have long been a supporter of introducing some version of test-early or test-first design into my intro programming courses, but I didn't find the unit testing ecosystem of C++ especially conducive to intro students' needs. So I wrote a new framework, called Unci (UNit testing with a Clean Interface---not the cleanest acronym but it does make for an attractive logo!).

Here's some of what Unci has to offer the intro student:

From the student's perspective, a ".u file" is a different kind of file that they can compile, just as they can compile a ".cpp file". Under the hood, the Unci source file is translated (i.e. compiled) into a valid C++ file, and compiled against the cppunit library. (At some point in the future, I hope to make it more standalone, but the cppunit dependence is invisible to the student.)

Runnable programs

The Unci system includes the following executables:

unci-translate
This is the core executable; it reads, from stdin or from a given filename, a file presumed to be in Unci format, and it produces, on stdout, a C++ source file suitable for compilation with a C++11-compliant compiler. This executable should normally be wrapped in some sort of script.
uncic
This script gets a filename ending in .u, does some lightweight error checking, and calls unci-translate and g++ to create a .o file suitable for linking the cppunit library and whatever is being tested. This command is sort of the moral equivalent of "g++ -c". If you have your students using Makefiles or some other build system, this script would probably be the one you'd use (by writing a rule that uses it to generate .o files from .u files).
compile
This script is a somewhat more fully-featured shell script that can be your students' one-stop shop for compiling code if they're using the command-line compiler commands. It can be used in place of "g++", making .cpp and .u files both effectively first-class "source code files" as far as the students are concerned. (It supports the "-c" option if they want to be producing .o files, and also: passes all other options to g++; prints "Success!" rather than succeeding silently; prevents them compiling .h files accidentally; turns on debugging (-g); turns on warnings; prints only the first error message to prevent spamming with hundreds of lines of error messages. Most of that has nothing to do with Unci but is useful for the same intro audiences that Unci targets.)

Other inclusions

A few other files come in handy if you're in a command-line development environment:

Vim support:
ftdetect-unci.vim
To be dropped into a ftdetect/ directory inside ~/.vim or $VIM: Recognises .u as the extension for filetype unci
ftplugin-unci.vim
To be dropped into a ftplugin/ directory inside ~/.vim or $VIM: Tells Vim to basically treat this like a C++ file
syntax-unci.vim
To be dropped into a syntax/ directory inside ~/.vim or $VIM: Tells Vim how to interpret the Unci syntax (keywords, etc)
Enscript support:
unci.st
To be added to enscript's hl/ directory (possibly inside /usr/share/enscript/) to make the bolding and such work for .u files. (As with vim syntax, mostly follows existing C++ rules, with a few tweaks.) Also requires addidion of the line 
/\.u$/    unci;
to the namerules section of enscript.st

Presentations

System requirements and download

Currently, this is squarely targeted at a linux-y environment, and if you want to rebuild it on your system, you'll need make, flex, and a willingness to tinker (or email me, and I can try to help). Once built, you'll need to be able to copy an executable and two scripts to /usr/local/bin or the equivalent, and have a compiler that handles C++11 code. (The scripts are written assuming g++ but are trivially tweakable to use clang++, or probably others.) Also, for now, the cppunit library has to be installed (sorry).

If you download it, drop me an email and I'll update you when I get my act together and update it.

Don Blaheta / blahetadp@longwood.edu