Last Update Time-stamp: "97/08/08 17:20:31 zdu"

For documentation on using the package, see the comments at the beginning
of Dim.h.

For documentation on generating Dim.h from Dim.hm4 see the comments at the
beginning of Dim.hm4.

CURRENT STATUS:

The package should currently be regarded merely as "proof-of-concept."  For
it to be practical, the compiler used should be much more stable than the
one I am currently using (g++) and generate better code.

Versions of this package have been compiled under gcc 2.5.8, gcc 2.6.0 and
gcc 2.6.3, 2.7.2.  Unfortunately, this package continues to stress g++.

Under gcc 2.6.3 the slide example compiles.  However the ratPow and period
examples (with DIM_NO_CHECK defined) give an internal compiler error.  On
the slideBug.cc example (which is slide.cc with dimensional errors) gcc
2.6.3 signals the first dimensional error, and then gets "confused and bails
out."  gcc 2.5.8 used to find all dimensional errors.

Under gcc 2.7.2 the slide example compiles.  On the slideBug.cc example
(which is slide.cc with dimensional errors) gcc 2.7.2 signals the first
dimensional error, and then gets "confused and bails out."  gcc 2.5.8 used
to find all dimensional errors.  ratPow and period examples (with
DIM_NO_CHECK and DIM_GPP_BUG defined) appear to work.

COMPILATION:

To compile the output routines for subsequent linking do:

gcc -c dimOut.cc

To compile the example program (for example slide.cc) use the command:

gcc slide.cc dimOut.o -lg++ -lm -o slide

Similarly for slideBug.cc.  For ratPow.cc and period.cc, see the comments
at the start of those files.

NOTES:

1.  I believe that the proposed C++ standard removes the restriction imposed
by the ARM which did not allow int template-arguments in function templates.
Hence the idea behind the Sigplan paper conforms FULLY with my understanding
of the proposed C++ standard. 

2.  After publication of the Sigplan paper, I received email from Lee
Nackman (lrn@watson.ibm.com) informing me that essentially the same ideas
are used in the book "Scientific and Engineering C++: An Introduction with
Advanced Techniques and Examples," by John J. Barton and Lee R. Nackman,
Addison-Wesley, 1994, ISBN 0-201-53393-6.  He also told me that the
problematic feature referred to in (1) above was accepted by the IBM CSet++
compilers for OS/2 and AIX.

FEEDBACK:

Please direct all bug-reports or comments to zdu@acm.org.