[ESS] ESS 25.1.0 listed as 'obsolete' when upgrading from MELPA?

[Martin Maechler]

>>>>> Mark Seeto via ESS-help 
>>>>>     on Mon, 27 Jan 2025 05:48:33 +1100 writes:

    > Hi Kevin,
    >> Any ideas of what I'm doing wrong, or how to fix things? Thanks for any
    >> advice or guidance. Using Emacs 29.3.

    > I'll start by saying that I'm not particularly knowledgeable in ESS or
    > Emacs, and my educated guess below could be partially or completely
    > incorrect.

    > I don't think you're doing anything wrong. I think
    > package-list-packages shows ess 25.1.0 as obsolete because
    > package-list-packages sees ess 25.1.0 as being an older version than
    > ess 20230419.1527. And I think it sees 25.1.0 as older because it
    > compares version numbers rather than dates, and the integer 20230419
    > is greater than the integer 25.

    > I looked in package.el to try to find where "obsolete" is defined, but
    > I didn't see it there. However, package.el uses comparisons of
    > versions. It uses the function version-to-list to convert a version
    > string to a list of integers, e.g. "25.1.0" to (25, 1, 0). And it uses
    > the functions version-list-<, version-list-<=, and version-list-= to
    > compare versions.

    > If I'm correct, your question is more about Emacs than about ESS.

    > See
    > C-h f version-to-list
    > C-h f version-list-<
    > etc.

    > Good luck with the upgrade,
    > Mark

Thank you, Mark.

.. but when we released  ess 24.01.0  and 24.01.1  last year,
these were *never* listed obsolete AFAIK,
but your explanation would lead to those, too.

But then, I do *not* see what you see, if I do

  M-x package-list-packages [Enter]
  and search (C-s) for  ' ess '

I see I have installed 24.1.1 and 25.1.0 is available ..
After I install 25.1.0, I see that and if I "click" [Enter]
I get the buffer explaining its "data", ending with

  Other versions: 24.1.1 (installed), 25.1.0 (melpa-stable), 20250110.1437 (melpa), 25.1.0 (gnu).

(and then I see 24.1.1 is now obsolete .. and I mark and
 uninstall it).

Martin


--
Martin Maechler
ETH Zurich, R Core and ESS core teams