I too agree with the pragmatic problems with the opacity that have
been presented.  Although I do think that there can be good reasons
for needing security, the oversimplified opacity flag is no actual
provision for security.  If anything is to be provided to this effect,
it would make no sense to have anything less than general capability
security, and I doubt whether anyone feels this would fall under the
scope of this SRFI.

Also, are sealed records at all useful in a single-inheritance system?
Are there any compiler optimizations permitted by them that would be
impossible otherwise?  I can't find any rationale in the document.