Per Tom comments, attached patch does following:
- Binary I/O support
- Use ActiveSnapshot instead of SerializableSnapshot
- Clarify function purposes
- remove dev targets from Makefile
There should not be any code depending on SerializableSnapshot,
if there is thats a bug that needs fixing. Fix is easy - use
SERIALIZABLE transaction isolation level.
Also - either Tom or Jan has changed the code to allow
xmin == xmax. Is such case really valid? In _recv
code I followed the style, but I was just wondering.
pgsql-patches by date
|Next:||From: Gregory Stark||Date: 2007-10-11 13:42:47|
|Subject: Re: [PATCHES] Eliminate more detoast copies for packed varlenas|
|Previous:||From: Heikki Linnakangas||Date: 2007-10-11 08:36:22|
|Subject: Re: Including Snapshot Info with Indexes|