On Wed, Jan 02, 2008 at 01:17:03PM +, Simon Riggs wrote:
> That's right; shame I forgot this before I started the thread...
Actually, I think your idea has merit, it's just not as easy as
originally thought. All splits, including multiple-level splits, can be
described as a sequence of "split
On Jan 2, 2008 6:24 PM, Martijn van Oosterhout <[EMAIL PROTECTED]> wrote:
> On Wed, Jan 02, 2008 at 04:46:11PM +0530, Gokulakannan Somasundaram wrote:
> > > All indexes are done by user-defined functions, even b-trees. People
> can
> > > make their own b-tree indexes by defining an operator class.
On Wed, 2008-01-02 at 13:54 +0100, Martijn van Oosterhout wrote:
> On Wed, Jan 02, 2008 at 04:46:11PM +0530, Gokulakannan Somasundaram wrote:
> > > All indexes are done by user-defined functions, even b-trees. People can
> > > make their own b-tree indexes by defining an operator class. Note that
>
On Wed, Jan 02, 2008 at 04:46:11PM +0530, Gokulakannan Somasundaram wrote:
> > All indexes are done by user-defined functions, even b-trees. People can
> > make their own b-tree indexes by defining an operator class. Note that
> > "user-defined" is this case means anything called via the fmgr
> > i
>
>
> All indexes are done by user-defined functions, even b-trees. People can
> make their own b-tree indexes by defining an operator class. Note that
> "user-defined" is this case means anything called via the fmgr
> interface.
Again, i think i have one more wrong understanding. My understandin
On Wed, Jan 02, 2008 at 04:04:48PM +0530, Gokulakannan Somasundaram wrote:
> On Jan 2, 2008 3:35 PM, Martijn van Oosterhout <[EMAIL PROTECTED]> wrote:
> > If the goal is to only store the insert, then we need to determine
> > during recovery which page the record needs to be added to. To do this
>
On Jan 2, 2008 3:35 PM, Martijn van Oosterhout <[EMAIL PROTECTED]> wrote:
>
> If the goal is to only store the insert, then we need to determine
> during recovery which page the record needs to be added to. To do this
> you need to go through the index, which can only be done by calling
> user-def
On Wed, Jan 02, 2008 at 02:49:35PM +0530, Gokulakannan Somasundaram wrote:
> On Jan 2, 2008 2:25 AM, Simon Riggs <[EMAIL PROTECTED]> wrote:
> > On Tue, 2008-01-01 at 14:02 -0500, Tom Lane wrote:
> > > How are you going to avoid the need to run user-defined functions
> > > (specifically, the btree c
On Jan 2, 2008 2:25 AM, Simon Riggs <[EMAIL PROTECTED]> wrote:
> On Tue, 2008-01-01 at 14:02 -0500, Tom Lane wrote:
> > Simon Riggs <[EMAIL PROTECTED]> writes:
> > > If we could log *only* the insert that caused the split, rather than
> the
> > > split itself, we would avoid that situation entirel
On Tue, 2008-01-01 at 22:20 +0100, Martijn van Oosterhout wrote:
> On Tue, Jan 01, 2008 at 08:55:58PM +, Simon Riggs wrote:
> > On Tue, 2008-01-01 at 14:02 -0500, Tom Lane wrote:
> > > Simon Riggs <[EMAIL PROTECTED]> writes:
> > > > If we could log *only* the insert that caused the split, rathe
On Tue, Jan 01, 2008 at 08:55:58PM +, Simon Riggs wrote:
> On Tue, 2008-01-01 at 14:02 -0500, Tom Lane wrote:
> > Simon Riggs <[EMAIL PROTECTED]> writes:
> > > If we could log *only* the insert that caused the split, rather than the
> > > split itself, we would avoid that situation entirely.
>
On Tue, 2008-01-01 at 14:02 -0500, Tom Lane wrote:
> Simon Riggs <[EMAIL PROTECTED]> writes:
> > If we could log *only* the insert that caused the split, rather than the
> > split itself, we would avoid that situation entirely.
>
> How are you going to avoid the need to run user-defined functions
Simon Riggs <[EMAIL PROTECTED]> writes:
> If we could log *only* the insert that caused the split, rather than the
> split itself, we would avoid that situation entirely.
How are you going to avoid the need to run user-defined functions
(specifically, the btree comparison functions) during replay?
When we split an index page we perform a multi-block operation that is
both fairly expensive and complex to reconstruct should we crash partway
through.
If we could log *only* the insert that caused the split, rather than the
split itself, we would avoid that situation entirely. This would then
me
14 matches
Mail list logo
