Re: Informix vs. Sybase vs. Oracle vs. (gasp) MS SQL Server

Jean-Marc van Leerdam wrote: > Anthony Mandic wrote: > > > Only if that lock was held for longer than it ought to have been. > > Regardless, I think one should NOT be troubled by things being done to > records that are of no concern to that user. With PLL you can (and > will) be troubled in such a way. Except in the case where one user has one row locked with RLL and another users want a summery of all rows including the locked one. Then you have to introduce further mechanisms to overcome this. But no lock, implies no problem and no further complexity. > I think, even in a set-based thinking environment, it should not be > the database that causes these problems. I strongly support the > availability of RLL (even if sometimes PLL would be a performance > advantage, I still want the option of RLL without cumbersome tricks or > fake identifiers). I agree. > > These should be fast. If there are a lot of them in one hit, I'd > > consider rethinking the approach used to design the app that > > does this. The implication here is that the app may be far too > > complex, thus being more vulnerable to problems. Keeping it > > simple never hurts. > > Just 'blaming' the apps complexity is no real solution too, I think we > should see the apps complexity and way of operating as a given. > Just because a RDBMS acts on sets and likes set-like operations, that > doesn't mean the customer does the same or should be trained to do so > (that sets us back some 30 years, when we still wanted to adapt the > users to the programs instead of the other way around). I think at this point I'd start to think of the meaning of relational. You could select one row initially and then sets of related rows. > IMHO a RDBMS should let the user operate on rows, completely > independent of the status and concurrent use of 'nearby' rows by other > users. > That implies RLL in my opinion. > > Also, in a set-based thinking environment: why is some element > 'closer' to another element than another element, and why does > modifying one element influence the success of modifications to SOME > other elements but not to ALL other elements? > Elements in a set should IMO be independent objects. I'd think relational again. Thus elements in a set are independant but related. -am