On Tue, 2002-07-09 at 17:49, Oliver Elphick wrote:
> On Tue, 2002-07-09 at 16:41, Hannu Krosing wrote:
> > On Tue, 2002-07-09 at 13:48, Oliver Elphick wrote:
> > > On Tue, 2002-07-09 at 01:30, Matthew T. O'Connor wrote:
> > > > Example: When PG 7.3 is released, the RPM / deb / setup.exe include the
> > > > postmaster binary for v 7.2 (perhaps two or three older versions...).
> > >
> > > That isn't usable for Debian. A package must be buildable from source;
> > > so I would have to include separate (though possibly cut-down) source
> > > for n previous packages. It's a horrid prospect and a dreadful kludge
> > > of a solution - a maintainer's nightmare.
> >
> > The old postmaster should not be built/distributed. As it is for
> > _upgrading_ only, you just have to _keep_ it when doing an upgrade, not
> > build a new "old" one ;)
>
> No, it doesn't work like that. You cannot rely on anything's being left
> from an old distribution; apt is quite likely to delete it altogether
> before installing the new version (to enable dependencies to be
> satisfied). At present I have the preremoval script copy the old
> binaries to a special location in case they will be needed, but that
> fails if the version is very old (and doesn't contain that code), and
> it's a very fragile mechanism.
>
> I never have understood why the basic table structure changes so much
> that it can't be read; just what is involved in getting the ability to
> read old versions?
The big change was from 6.x to 7.x where a chunk of data moved from end
of page to start of page and tableoid column was added. Otherways the
table structure is quite simple. The difficulties with user _data_ can
be mainly because of binary format changes for some types and such.
But I still can't see how will having a binary dumper that does mostly
the work of [ old_backend -c "COPY tablex TO STDOUT" ] help us here.
IIRC the main difficulties in upgrading have always been elsewhere, like
migrating always changing system table data.
----------
Hannu