Bug#1039105: Fails to start with Emacs 28

[Ian Jackson]

Control: severity -1 serious

This bug makes the package very hard to use.  It fails to work, by
default.

-- 
Ian JacksonThese opinions are my own.  

Pronouns: they/he.  If I emailed you from @fyvzl.net or @evade.org.uk,
that is a private address which bypasses my fierce spamfilter.

Bug#1039105: Fails to start with Emacs 28

[Ian Jackson]

Dirk Eddelbuettel writes ("Bug#1039105: Fails to start with Emacs 28"):
>   After updating my main machine (and the only one running vm along and
> exim4, dovecot, spamassassin and whatnot) to Ubuntu 23.04 with its Emacs 28.2
> (in an upgrade from 22.10 with Emacs 27.*), I found that vm (which I have
> been using all those years, still with some hooks and key-binding you emailed
> to in the mid-1990s (!!)) would no longer start.

Thanks for the report.  And thanks to the others for the further
information.

I can reproduce the problem and the suppression of byte compilation
does indeed fix it.

It's a bit funky to reproduce, because emacs likes to byte-compile
things when it's idle with the effect that naive "emacs --batch" runes
don't show the problem.

I'll do a stable update.  But first I am going to upload to sid,
as per the stable release policy.  And I'm adding a test, since I did
manage to get a reliable repro.

I confess I don't understand the upstream package very well.

Ian.

-- 
Ian JacksonThese opinions are my own.  

Pronouns: they/he.  If I emailed you from @fyvzl.net or @evade.org.uk,
that is a private address which bypasses my fierce spamfilter.

Bug#1039105: Fails to start with Emacs 28

[Dirk Eddelbuettel]

Package: vm
Version: 8.2.0b-8
Severity: normal

Ian,

  After updating my main machine (and the only one running vm along and
exim4, dovecot, spamassassin and whatnot) to Ubuntu 23.04 with its Emacs 28.2
(in an upgrade from 22.10 with Emacs 27.*), I found that vm (which I have
been using all those years, still with some hooks and key-binding you emailed
to in the mid-1990s (!!)) would no longer start.

The error was indicative of elc compilation woes. I found a post by the magit
author suggesting removal and reinstallation (of his package). That worked,
so I was relieved, but it worked only once.

Looking around, I stumbled across the Fedora page for vm. It contains a
reference to a change in this commit

  
https://src.fedoraproject.org/rpms/emacs-vm/c/909b0bc357976252c51502bf17ed1efc6aeb7b97?branch=rawhide

which boils down to

;; Startup settings for VM
  + ;;
  + ;; For some reason, native compilation breaks VM. As a workaround until the
  + ;; problem is understood and fixed, disable native compilation of all VM
  + ;; lisp files.
  + (eval-after-load "comp"
  + '(if (boundp 'native-comp-deferred-compilation-deny-list)
  + (add-to-list 'native-comp-deferred-compilation-deny-list 
"/vm.*\.el")))

I added this in a local -9 build to the top of debian/vm.emacsen-startup
which now has

  ;-*-emacs-lisp-*-
  ; arch-tag: 045640fd-0ff2-45b7-a29f-68e4b8378fbf

  ;; Borrowed with thanks from Fedora at
  ;; 
https://src.fedoraproject.org/rpms/emacs-vm/c/909b0bc357976252c51502bf17ed1efc6aeb7b97?branch=rawhide
  ;;
  ;; For some reason, native compilation breaks VM. As a workaround until the
  ;; problem is understood and fixed, disable native compilation of all VM
  ;; lisp files.
  (eval-after-load "comp"
  '(if (boundp 'native-comp-deferred-compilation-deny-list)
  (add-to-list 'native-comp-deferred-compilation-deny-list 
"/vm.*\.el")))

It would be nice if this could be applied to the package as a hot-fix as
failing to start is a little off-putting. We could probably also skip the
installation on start but there is enough going on in that installer script
that I left things were they were.

Happy to locally refinements or variants if you have any as I obviously have
a test bed for the bug.

Cheers, Dirk

--
dirk.eddelbuettel.com | @eddelbuettel | e...@debian.org