Postgres Pro
Downloads
Home   >   mailing lists

Thread: The new FE/BE protocol

The new FE/BE protocol

From
Bruce Badger
Date:
I read that there is a significant new version of the FE/BE protocol on
the way in the next version of PostgreSQL.

I'm the author of the Smalltalk PostgreSQL driver, and I'd like to make
sure that support for the new protocol is available as soon as possible.

Where will the new protocol be documented (& will documentation precede
coding?)?

At what point should I start making dev-builds of PostgreSQL to test out
the new protocol? (A pointer to the docs on how to do that would be nice
too).

Many thanks,Bruce

Re: The new FE/BE protocol

From
ljb
Date:
bruce_badger@badgerse.com wrote:
> I read that there is a significant new version of the FE/BE protocol on
> the way in the next version of PostgreSQL.
> 
> I'm the author of the Smalltalk PostgreSQL driver, and I'd like to make
> sure that support for the new protocol is available as soon as possible.
> 
> Where will the new protocol be documented (& will documentation precede
> coding?)?
> 
> At what point should I start making dev-builds of PostgreSQL to test out
> the new protocol? (A pointer to the docs on how to do that would be nice
> too).

I second this, as my Tcl driver also depends on the protocol. Can we also
be reassured about compatibility (e.g. new backend will still talk to old
clients)?

Re: The new FE/BE protocol

From
Tom Lane
Date:
Bruce Badger <bruce_badger@badgerse.com> writes:
> Where will the new protocol be documented (& will documentation precede
> coding?)?

I'd like to have a revised protocol spec out within the next few days
--- but it'd be subject to change as the implementation proceeds.

> At what point should I start making dev-builds of PostgreSQL to test out
> the new protocol?

Probably not much point until the protocol changes are all implemented,
which'd be, say, late this month.
        regards, tom lane

Re: The new FE/BE protocol

From
Tom Lane
Date:
ljb <lbayuk@mindspring.com> writes:
> I second this, as my Tcl driver also depends on the protocol. Can we also
> be reassured about compatibility (e.g. new backend will still talk to old
> clients)?

Yes.

The contents (layout and spelling) of error messages are likely to
change, which could break frontends that test for specific messages
--- but I'd opine any such frontends are already broken by message
internationalization.  Otherwise, an old client should see no change.
        regards, tom lane