Certainly the Sybex OCP Study Guides for 8i had no reference to the performance implications of cursors.  I know, because I searched through them...  I also grabbed every Oracle reference book owned by anyone in the entire department, most of which came from Oracle Press, and which included "tuning and diagnostic" books; while all of them dealt with HOW to code cursors, NONE of them so much as hinted that there were performance implications to doing so.  Further, every one of the dozen or so developers and DBA's that I discussed the subject with, were under the "impression" that Oracle was somehow different, and thus immune from the performance problems that SQL Server labored under.  I was assured time and again that I should quit barking up that tree.  It wasn't until I devised several tests and proved my point that anyone would listen. 

The problem, like you point out, is that those who are used to procedural (or OO) code are uncomfortable with set-based operations and they tend to see cursors as very familiar and "easy."

Granted, my experience is subjective and not in any way a comprehensive study of the subject, and I have run into Oracle DBA's and developers who know better.  However, the vast majority of SQL Server DBA's/developers are aware of the drawbacks of using cursors because, among others, Microsoft is very up-front about their potential problems.  At least a majority of people I have discussed the issue with (both SQL Server and Oracle folks) labor under the misconception that somehow Oracle "is optimized for cursors" or some such claptrap. 

Don't misunderstand, I'm not bagging Oracle, I've worked with it for several years now and appreciate it's strengths, but I'm also not blind to its weaknesses...