No, Erik, they're just making it difficult/impossible to evaluate their product.
disagree, see below

Nobody has time to spend on evaluating a handful of compilers if they have to make a living too.
well, if you do not have the time to evaluate "a handful", your points are moot


The result is that a lot of folks buy the most expensive toolset because someone else is paying for it, or a cheaper and sometimes less adequate one because they're not. Once they've bought it, they realize they've made a mistake, but become an advocate for it because they have to obfuscate their error.
if you want to see an example of the opposite, search for my comments on the SILabs IDE which I use extensively. Also, a good source is the products forum, one compilers forum has as a response to every third entry "will be fixed in next release" do you REALLY need to evaluate that one

I'd like to have time to fiddle with various compiler evaluations, but just can't take the time. Expecting me to do what I have to do while evaluating someone's good or bad calendar-limited product evaluation version is unrealistic.
who says 'calendar limited, if you are "known honest" any manufacturer (correction, last evaluation time one of eight would not *1) will loan you the software without restrictions.

Software is imperfect, and, in fact, probably more so than anything else, except, perhaps, government. One has to have time to see whether the warts on this product are more tolerable than those on that one. From what I've observed, even the best software houses have regularly sacrificed quality for schedule. It's like what our gov is doing to the healthcare debate. Eventually, it becomes, "Do something even if it's wrong!"
I have yet to foind a fault in the essential part of the keil tools (compiler, assembler and linker) And I do not give a shit about the rest (optimizer, editor, device tables).

Erik

*1 evaluation refused with the following comment "you do not need to evaluate, ours is the best", guess my belief in that one.