bug#62750: 29.0.50; Commands 'package-update' and 'package-update-all' should be called '*-upgrade'

[Eli Zaretskii]

> Date: Mon, 24 Apr 2023 21:54:58 +0300
> Cc: monnier@iro.umontreal.ca, philipk@posteo.net, larsi@gnus.org,
>  62750@debbugs.gnu.org
> From: Dmitry Gutov <dmitry@gutov.dev>
> 
> On 24/04/2023 20:28, Adam Porter wrote:
> > On 4/24/23 07:02, Eli Zaretskii wrote:
> > 
> >> Me, I have only one potential issue: since "update" just means "delete
> >> the installed version, then install another version", it could be
> >> easily made to downgrade, not just to upgrade.  So if we ever would
> >> like to allow downgrading, the new names will get in the way.  But if
> >> this is not an issue we should be bothered about, it's fine by me.
> > 
> > IMHO, a command to downgrade ought to be a separate command with a 
> > different name--not only to reduce confusion, but because downgrading 
> > packages is an operation that is more likely to require manual user 
> > intervention, such as recompiling other packages that depend on the 
> > downgraded package (e.g. if struct or macro definitions change, or 
> > symbols disappear).
> 
> That might also be the case when upgrading a package that some others 
> depend on (newer version could also have macros deleted or renamed).
> 
> Either way, though, we could make it a separate command.

A separate command that does the same, or almost the same, as
package-upgrade?  That's uneconomical, let alone elegant.

> Or even augment the current one: (package-upgrade 'name 
> "some-older-version") has a similar feel to (forward-char -1), not 
> exactly unfamiliar to us.

We are talking about invoking commands, not about Lisp programs.  How
many times did you do "C-- C-f" instead of "C-b"?

> We don't keep older versions around in ELPA anyway, so for now the
> question is moot.

I was trying to raise a possible future issue.  We all know that
command names, once they gain enough tenure, cannot be easily changed.
So this is the time to think about future issues; we won't have
another chance.  It's exactly why we should consider what is today a
"moot point" but could be a real one later.