pgsql: Added ECPGget_PGconn() function to ecpglib, courtesy of Mike

meskes@postgresql.org (Michael Meskes) writes:

Log Message:
-----------
Added ECPGget_PGconn() function to ecpglib, courtesy of Mike Aubury.
Bumped library version to 6.1.

Tags:
----
REL8_3_STABLE

This should absolutely *not* have been done in a stable release branch.
Kindly revert.

regards, tom lane

On Thu, Mar 20, 2008 at 04:36:15PM -0400, Tom Lane wrote:

This should absolutely *not* have been done in a stable release branch.

I thought about this for a while before committing but failed to see
where an additional function could harm the stable release. There is
however an advantage, namely enabling aubit with PGSQL 8.3 again.
Without this change it won't work with this release.

I do agree though that normally you don't want to do this in a stable
branch.

Kindly revert.

Is this consensus around here? I don't to argue about this just tried to
help some people.

Michael
--
Michael Meskes
Email: Michael at Fam-Meskes dot De, Michael at Meskes dot (De|Com|Net|Org)
ICQ: 179140304, AIM/Yahoo: michaelmeskes, Jabber: meskes@jabber.org
Go VfL Borussia! Go SF 49ers! Use Debian GNU/Linux! Use PostgreSQL!

Michael Meskes <meskes@postgresql.org> writes:

On Thu, Mar 20, 2008 at 04:36:15PM -0400, Tom Lane wrote:

This should absolutely *not* have been done in a stable release branch.

I thought about this for a while before committing but failed to see
where an additional function could harm the stable release. There is
however an advantage, namely enabling aubit with PGSQL 8.3 again.
Without this change it won't work with this release.

It's not the extra function that I object to so much as the library
version number bump. That's going to create problems for packagers.

regards, tom lane

On Fri, Mar 21, 2008 at 10:30:18AM -0400, Tom Lane wrote:

It's not the extra function that I object to so much as the library
version number bump. That's going to create problems for packagers.

Ah, I didn't think far enough. Sorry, your definitely right on this one.
Reverted.

Michael
--
Michael Meskes
Email: Michael at Fam-Meskes dot De, Michael at Meskes dot (De|Com|Net|Org)
ICQ: 179140304, AIM/Yahoo: michaelmeskes, Jabber: meskes@jabber.org
Go VfL Borussia! Go SF 49ers! Use Debian GNU/Linux! Use PostgreSQL!

Michael Meskes wrote:

On Fri, Mar 21, 2008 at 10:30:18AM -0400, Tom Lane wrote:

It's not the extra function that I object to so much as the library
version number bump. That's going to create problems for packagers.

Ah, I didn't think far enough. Sorry, your definitely right on this one.
Reverted.

But is it OK to add a function to the library without bumping the
version? AFAICS all you have reverted is the version bump, so we could
have two versions of the library with the same version number, one with
the new function and one without - that seems somewhat undesirable.

Also, regardless of the safety or otherwise, isn't it our standard
practice not to change anything in stable releases except for bug fixes?
That's part of what makes Postgres so stable, in fact.

cheers

andrew

On Tue, Mar 25, 2008 at 05:34:44PM -0400, Andrew Dunstan wrote:

But is it OK to add a function to the library without bumping the
version? AFAICS all you have reverted is the version bump, so we could

With most people refering to ecpg 8.3.1 (for instance) anyway I think
the risk is rather small.

have two versions of the library with the same version number, one with
the new function and one without - that seems somewhat undesirable.

It is indeed.

Also, regardless of the safety or otherwise, isn't it our standard
practice not to change anything in stable releases except for bug fixes?
That's part of what makes Postgres so stable, in fact.

I still fail to see how this added function could pose a problem. It
does, however, fix a bug - make that a missing, undocumenterd feature -
that causes the aubit application to break.

Michael
--
Michael Meskes
Email: Michael at Fam-Meskes dot De, Michael at Meskes dot (De|Com|Net|Org)
ICQ: 179140304, AIM/Yahoo: michaelmeskes, Jabber: meskes@jabber.org
Go VfL Borussia! Go SF 49ers! Use Debian GNU/Linux! Use PostgreSQL!

Michael Meskes wrote:

Also, regardless of the safety or otherwise, isn't it our standard
practice not to change anything in stable releases except for bug fixes?
That's part of what makes Postgres so stable, in fact.

I still fail to see how this added function could pose a problem. It
does, however, fix a bug - make that a missing, undocumenterd feature -
that causes the aubit application to break.

Arguably, it is aubit that is/was buggy in relying on the undocumented
"feature".

I don't think this particular change poses a great danger, but I do
think it's a bad precedent.

Anyway, I've said my piece, and don't want to take it further.

cheers

andrew