Re: New hooks for snapshot and transaction ID acquisition from external source - Mailing list pgsql-hackers

From Tom Lane
Subject Re: New hooks for snapshot and transaction ID acquisition from external source
Date
Msg-id 25426.1361617614@sss.pgh.pa.us
Whole thread Raw
In response to New hooks for snapshot and transaction ID acquisition from external source  (Michael Paquier <michael.paquier@gmail.com>)
List pgsql-hackers
Michael Paquier <michael.paquier@gmail.com> writes:
> What do you think about adding hooks for transaction ID and snapshot
> acquisition?

Not much.  Those places are not only hot spots but remarkably delicate.
It's hard to see much that a plugin could do there that wouldn't be
pretty deleterious to both performance and reliability.  As an example,
throwing an error in either place is dangerous, and taking any locks
they don't already take even more so.

Now of course, if your plugin breaks your database that's maybe not my
problem, but I'm concerned that even just adding a test to see if
there's a hook to call could result in measurable slowdown.
        regards, tom lane



pgsql-hackers by date:

Previous
From: Tom Lane
Date:
Subject: Re: bugfix: --echo-hidden is not supported by \sf statements
Next
From: Pavel Stehule
Date:
Subject: Re: bugfix: --echo-hidden is not supported by \sf statements