On Fri, 27 Oct 2006 09:21:14 -0400, Craddock, Chris <[EMAIL PROTECTED]> wrote:
>... >I don't think he did. If you widen the definition of reentrant to >include all external data then all bets are off. I would argue that such >a program is just badly designed no matter what purported attributes it >has. The system provides well-behaved mechanisms for well-behaved >programs. It is a bit much to expect miracles when the program violates >serialization rules. >... I admit I was one of those gullible enough to believe the myth (which Tom Marchant has exposed as bogus) that RENT ever meant anything other than "not self-modifying". It apparently never had anything to do with whether the code could safely be reentered. So I would change your statement to "If you widen the definition of reentrant to mean reenterable, all bets are off". That is really the important issue, but it's impossible to determine programmatically. The RENT attribute checked by the Assembler is a lot closer to a REFR test (as has already been stated in this thread, I think) and is a very rough test at that, prone to both false positives and false negatives. Without knowledge of those "well-behaved mechanisms for well-behaved programs" you mention there is no way to determine reenterability. Pat O'Keefe ---------------------------------------------------------------------- For IBM-MAIN subscribe / signoff / archive access instructions, send email to [EMAIL PROTECTED] with the message: GET IBM-MAIN INFO Search the archives at http://bama.ua.edu/archives/ibm-main.html
