With a car, you have repeated visits to the various dealers lots, and can take your time, since the individual car is not what you'll get to evaluate anyway. Last time I bought a car, it took me two years to do the job. I ended up with a newer version of what I'd been driving for the past 20+ years, but I did look at a number of other vendors' products. Likewise, I can get some "feel" from software vendors' "evaluation" version, but it's mainly the doc's I'm interested in. It's nearly impossible to make a comparison/contrast effort produce anything reasonable when comparing widely divergent "eval" packages. The doc is generally so thoroughly contaminated with marketing drivel that it's not of much use.

Most importantly, however, calendar-based limits on how long an eval version is available to the prospective purchaser of a piece of code makes evaluation impossible. Evaluation, if you're not being paid strictly to perform an evaluation, requires time, and you have time in bits and pieces, as your "real" work allows. In those cases where a vendor has provided a time-limited version, it has provided enough time to read the accompanying doc's, but little else. One can, of course, run the demo pieces, but those are pre-engineered to show the product in favorable light. In the end, you're flying blind.

As I previously mentioned, that scheme of counting the days on which the package is opened is probably the most satisfactory approach. It's a pity that such an approach hasn't been used by software vendors.

RE