STCKCONV should not be altered; it should do what it has always done. Too much existing code woulfd be broken by changing its behavior now.
Another facility, call it say STCK2UTC, could usefully be defined on top of it; but that is another matter. Maintenance for a table of STCKE values and their corresponding cumulative leap-second counts is unproblematic; The IERS provides, minimally, six months' notice of impending leap-second corrections; and automated BIPM notifications of these corrections are also available. I have a literally exhaustively tested, much used PL/I procedure that does GLB-seeking binary search in a self-defining STCKE table available; and I should be happy to send a copy of the source code for it to anyone who wants it. (It could be converted to C or another C-like language without much difficulty. Its chief "limitation" is that it deals only in STCKE, not STCK values.) John Gilmore, Ashland, MA 01721 - USA
