Re: bounded quantification and record-update problems

• To: types@theory.LCS.MIT.EDU
• Subject: Re: bounded quantification and record-update problems
• From: Hans Boehm <boehm@rice.edu>
• Date: Thu, 20 Oct 88 15:13:14 EDT
• Sender: meyer@theory.LCS.MIT.EDU

Date: Thu, 20 Oct 88 11:00:46 CDT
To: rdt%QUCIS.BITNET@mitvma.mit.edu, types@theory.LCS.MIT.EDU

It seems to me that one of the underlying problems here is simply
that subtyping relationships between variables or references are
not the same as for simple values.  It may be possible to treat an
integer value as a real value in all cases, but the same is definitely
not true for REFERENCES to integers.  (The required inclusion is
reversed on the left side of an assignment.) Similarly, it appears to me
that the subtyping rule for assignable records must be different
>from that for nonassignable ones.  (The fact that fields are subtypes
of the corresponding fields does not imply that a subtype relationship
holds between assignable record types.)

Hans-J. Boehm
boehm@rice.edu

• Prev: bounded quantification and record-update problems
• Next: the Robinson-Tennent coercion
• Index(es):
  • Main
  • Thread