With Bill's permission, I'm forwarding his review to the list. -Fred ---------- Forwarded message ---------- From: Bill Myers <wlmyers at stellarscience dot com> Date: Jun 6, 2006 10:14 AM Subject: Review of Andy Little's Physical Quantities System To: fred.bertsch@gmail.com Cc: Andy Little <andy@servocomm.freeserve.co.uk> == Summary: recommend accept. Please! :-) == Design: My perspective is a component user; I have little expertise in constructing boost-level-of-sophistication components. The user API is fine. == Implementation: I've compiled and run all the pqs example programs. They compile with no warnings and execute correctly. I've added a user-defined t1_quantity, fluence. I did this before documentation was available; Andy was cheerfully helpful in walking me through the steps, but it's daunting for the casual user (as Andy notes in the documentation). == Documentation: The PDF version of the document has considerable content, with a few editorial issues (typography, spelling, and such). I spent a couple of hours reading through all of it; I found it no more dense than existing boost component documentation, and certainly more up-to-date than some of them (unit test). == Usefulness: PQS fills a critical need for which there is no alternative C++ solution. Even though some may regard the present implementation as incomplete (Andy himself notes important TODOs), I believe that it is very useful as it now stands, and I need a "stamp of approval" so that I can begin introducing it into our scientific analysis projects. == Background: My platform is Apple (gcc 4.0.1, Mac OS X 10.4.6). I'm using boost 1_33_1 obtained via darwinports. My coworkers are pretty much all on WindowsXP with VisualStudio -- anything I create has to work on that platform; they are also using boost 1_33_1. I've been experimenting with pqs since January 2005, although none of our code can depend on pqs until it becomes a formal boost library. Bill Myers