L' Eomot Inversé (10/19/2012)
Maybe some programmers, but other programmers consider the statement
that SQL Server doesn't correctly manage the size of tempdb to be either incorrect or at least uncertain, so we don't
consider it to be a bug.
Its a new heads up to me, I'm still digesting it and could be misinterpretting the whole situation. Here was the first heads up for me, posted by Sean Lange:
which refers to
which coming direct from Microsoft does give some legitimacy to the issue. I may have spoken too soon about it "not being considered a bug" though as books on line (on my install) gives:
"The database being shrunk does not have to be in single user mode; other users can be working in the database when it is shrunk. This includes system databases."
so given the conflicting info, maybe the situation is not quite "not a bug", but I certainly am at least now aware of the issue.
Ah. I think I misunderstood you.
If you had said "shrink" (implying either shrinkdatabase or shrinkfile) I would have agreed that was bad (and I do agree MS have a bug there). In my opinion they do have a bug there - but they have clearly admitted that doing this can cause corruption of tempdb, so I don't think they are denying it. Whether shrink (as opposed to alter database modify) should be considered a legitimate method for the user to reduce the size of tempdb without ensuring a quiesced system by going into single user mode is neither here nor there (lets be clear: I don't consider it a legitimate method at all): until the documentation says you can't do that and the user interface prevents you from doing it, either they fix it so that it wroks reliably or they have a bug.
When you referred to SQL Server not managing the size of tempdb correctly, I assumed you were complaining about SQL Server not managing the size correctly - but actually you are referring to a bug that happens when the user (not SQL Server) attempts to manage the size of tempdb instead of letting SQL Server carry out its own management, guided by required start sizes, growth rates, and max sizes specified by the user. So you weren't assertuing a bug in the part that I thought you were.
Hence the misunderstanding.