Re: A proposal to change psqlodbc

Hiroshi Inoue <Inoue(at)tpf(dot)co(dot)jp> writes:
> I can wait to commit it if the 7.2 branch is near
> but have a basic question.

> What does 7.2 development mean for clients(psqlodbc) ?

You're correct that as far as ODBC's development goes, there's no reason
to tie it to server development cycles. But *as long as it's part of
the Postgres source tree* it needs to respect the Postgres cycle.

In particular, it won't do to risk introducing development-type bugs
into ODBC just before 7.1.1 release, because that release is going to go
out with only minimal beta testing. Major changes that are not forced
by serious bugs should happen on a development branch, not a
stable-release branch --- that's just plain common sense.

If this creates problems for ODBC, then maybe that needs to be a
separate project with a CVS tree somewhere else, and we'll go back
to just packaging a recent stable release of ODBC with Postgres
releases.

> P.S. The change would be a bug fix in a sense though
> I'm not sure. Both the driver manager and each ODBC
> driver have ODBC functions of the same name. ODBC
> function calls inside a driver may call the functions
> of the driver manager.

We've been working around that with -Bsymbolic on platforms where
it's needed. I agree that not naming the functions alike is a better
solution; but I don't think it's critical enough to justify a post-
release patch ...

regards, tom lane