Re: why do you apply undo before redo?

From: Ryan <>
Date: Sun, 18 Apr 2004 19:17:09 -0400
Message-ID: <O1Egc.5075$uF3.2623_at_lakeread04>

"Jan Hidders" <> wrote in message news:HFDgc.77172$
> Ryan wrote:
> > "Jan Hidders" <> wrote in message
> > news:anCgc.77108$
> >>Ryan wrote:
> >>>"Jan Hidders" <> wrote in message
> >>>news:Cwzgc.76768$
> >>>>Suppose that in the REDO list there is a transaction with an operation
> >>>>that changed a field from "yes" to "no", and in the UNDO list there is
> >>>>transaction with also an operation that changed the same field also
> >>>>"yes" to "no". Then what is the result of the field after REDO - UNDO
> >>>>and what after UNDO - REDO?
> >>>
> >>>I do not see why this matters. It does not matter what the state of the
> >>>tuple is during recovery, only when it is complete.
> >>
> >>That is what I'm talking about. What is the result of the field after
> >>recovery has completed in the two cases? The first case being first REDO
> >>and then UNDO and the second being the reverse.
> >
> > undo only takes place for uncomitted transactions. undo for committed
> > transactions are ignored. are you telling me the author is assuming that
> > undo is applied even for committed transactions?


> No, I'm not. In the example the transaction in the UNDO list did not yet
> commit, and the transaction in the REDO list did. Does the book not
> explain what the REDO and the UNDO list are and how they are constructed?

I know what undo and redo are. How is it possible to commit redo and not undo? Commits are on a transaction by transaction basis? >
> -- Jan Hidders
Received on Mon Apr 19 2004 - 01:17:09 CEST

Original text of this message