Re: ANSI vs. Non-ANSI

On 10 Jan 1995, Caelus wrote: > : >stuff snipped< > : BTW, isn't it a MUST to log out of your terminal when you are going for > : lunch, I mean security and all that :-) > > No, from a designers viewpoint st least SOME users are blithering idiots > who can be counted upon to use their ignorance to find the weakest point > in your design and wreak the most havoc thereby. Bingo, right to the point. I do have the pleasure of being one myself sometimes. :-) (both idiot and wracked designer). > > Yes, we do this as a matter of course AS LONG AS the use has nothing in > process (no modification or new entry in process, etc). We put the > timeout period in the user settings file so that it may be user specific, > but there is a global default. Shouldn't the database admonistrator have more say on such things?? I would rather set different time settings for different kinds of operations, rather than based on user. Its not like some users are that much slower than others or what. But some operations are critical, and need to be completed fast, while others can be more lenient. Just my humble opinion. > > A point that wasn't specifically brought out in this thread, but which we > consider to be gospel, is that the duration of locks must not be > dependent upon the whims of the users (a user should not be able to leave > the customer master entry locked while he/she goes to lunch). This means we > use optimistic locking and other schemes to accomplish updates and > other transactions. Really, the more I think of it, the crux is making sure that the user is really not doing anything. And, wouldn't the avoidance of statements like "SELECT .... FROM ..... FOR UPDATE" elminate such issues. Okay, that is a stupid suggestion. No, I wouldn't want to write the codes to replace it too. But, I do think that its worth considering during design and coding. Anyway, there is some stuff earlier about gracefully killing an application when it is idle for some time. I'm sure that could be modified to rollback a transaction after timeout. Afterall, what we really want is that other users can proceed with their work. =-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-= Calvin Ng email: <calvinng@technet.sg> --------------------------------------------------------------------------- [DISCLAIMER] Read at your own risks. The author is not liable for any consequences directly or indirectly involved with reading this mail. This mail MAY NOT be interpreted as those of my employer's. --------------------------------------------------------------------------- This mail is sent with the help of Chameleon TCP/IP for Windows. =-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=