Re: pg_validatebackup -> pg_verifybackup?

On Sun, Apr 12, 2020 at 5:22 PM Robert Haas <robertmhaas(at)gmail(dot)com> wrote:

> On Sun, Apr 12, 2020 at 11:02 AM Magnus Hagander <magnus(at)hagander(dot)net>
> wrote:
> > There's https://en.wikipedia.org/wiki/Pg_(Unix).
> >
> > So it's been removed from posix, but not unlikely to be around. For
> example, I see it on a server with Debian 9 (Stretch) or Ubuntu 16.04 which
> is still well in support (but not on a RedHat from the same era).
>
> Well, if it's around on older distros, but not in the newest versions,
> I think we should try to lay speedy claim to the name before something
> else does, because any other name we pick is going to be longer or
> less intuitive or, most likely, both. There's no guarantee that
> PostgreSQL 14 would even get packaged for older distros, anyway, or at
> least not by the OS provider.
>

It definitely won't be by the OS provider, however it will be by *us*. Our
apt and yum repositories support all our versions on "all supported"
versions of the upstream distros. So we should at least have a plan and a
story for how to deal with that, and make sure all our own packagers deal
with it the same way.

We could also have an alternate name, like pgsql, and make 'pg' a
> symlink to it that packagers can choose to omit. (I would prefer pgsql
> to pg_ctl, both because I think it's confusing to adopt the name of an
> existing tool as the meta-command and also because the underscore
> requires pressing two keys at once, which is slightly slower to type).
> But there is no way anyone who is a serious user is going to be happy
> with a five-character meta-command name that requires six key-presses
> to enter (cf. cvs, git, hg, yum, pip, apt, ...).
>

Agreed, pgsql would certainly be better than pg_ctl.

--
Magnus Hagander
Me: https://www.hagander.net/ <http://www.hagander.net/>
Work: https://www.redpill-linpro.com/ <http://www.redpill-linpro.com/>