>>I am using semaphore locking method,.every table I have I create a new field >thathat contains a logical field to determine if the record is locked or not...
>>simply disallow an edit if this field is .T.
>
>I have heard of the technique. However, I have never seen compelling evidence to resort to semaphores. If something crashes, you need to manually go and unlock records. Seems totally counter-productive to me.
>
I have heard of the technique also, SBT accounting software uses it. In the field it is a total pain as the end user does exit badly from time to time and I get these panic calls "I can't get in". They provide a utility to clear the flags but with the newer Lock()/Rlock() systems I agree the semaphore system is an out dated technique. I believe the only reason SBT used it was because they were "cruching" code written for Dbase II.
I use exclusively Lock()/RLock() with Begin Transaction/End Transaction. It seems to work fine and greatly reduces my programming overhead.
Good Luck
Terry
It is impossible to make programs idiot proof. Idiots are too cleaver.
MCP( Tcp/Ip )