Re: Synchronized Scan WIP patch

On Thu, 2007-03-22 at 16:43 -0400, Bruce Momjian wrote:
> Will use '16' rather than '100'.
>
> Your patch has been added to the PostgreSQL unapplied patches list at:
>
> http://momjian.postgresql.org/cgi-bin/pgpatches
>
> It will be applied as soon as one of the PostgreSQL committers reviews
> and approves it.
>

Here is the latest version, which includes the change to report every 16
pages.

This patch has the following improvements:

* reporting interval to 16 pages
* rearranges the scan location tracing output to work regardless of the
reporting interval. Previously it did not trace the output correctly if
the logging interval was not an even multiple of the reporting interval
* GUC trace_sync_seqscan=<bool> now controls whether the DEBUG output
is generated or not. If this is true, a lot of logging output will be
generated at DEBUG3.
* You can set sync_seqscan_threshold=<-1.0 ... 100.0>. Positive values
are treated as a fraction of NBuffers. Negative values disable sync
scans.

Still TODO:

* Publish my test results (I've collected much of the raw data already
on this version of the patch)
* SGML documentation (after we stabilize the GUC names and meanings)
* Possibly remove sync_seqscan_threshold=<real> and instead use a
simple enable/disable boolean that sets the threshold at a constant
fraction of NBuffers (most likely the same fraction as Simon's recycle
buffers patch)

Regards,
Jeff Davis