RE: More on SrPersist buffers


Subject: RE: More on SrPersist buffers
From: Paul Steckler (steck@ccs.neu.edu)
Date: Wed Dec 05 2001 - 13:45:43 EST


> Can you invalidate the "deallocated" ones? That is, have a wrapper
> structure with one extra field that indicates if the wrapped structure
> has been freed and when it is used with Scheme code, that flag could
> be
> checked and an error signalled?

I mentioned exactly that to Will at lunch as a possible solution.

The problem is, the ODBC calls which use the buffers wouldn't check for
the dealloc-bit. Those calls aren't passed buffers directly; there's
hidden state that associates buffers with result sets.

One could imagine keeping a table of buffers, and putting in checks in
the wrappers of all ODBC procedures that might use such buffers. That
duplicates some of the bookkeeping (name a word with three
double-letters in a row, ed.) done by the ODBC Driver Manager. It could
be done, though.

-- Paul



This archive was generated by hypermail 2b28 : Wed Dec 05 2001 - 13:46:12 EST