GNU bug report logs - #38324
27.0.50; package corruption? cannot de install packages

Previous Next

Full log

View this message in rfc822 format

From: Stefan Kangas <stefan <at> marxist.se>
To: Uwe Brauer <oub <at> mat.ucm.es>
Cc: 38324 <at> debbugs.gnu.org
Subject: bug#38324: 27.0.50; package corruption? cannot de install packages
Date: Sat, 25 Jan 2020 06:01:39 +0100

Uwe Brauer <oub <at> mat.ucm.es> writes:

> Steps to reproduce.
>
>     1. I start /opt/emacs27/bin/emacs -q
>
>     2. M-x package-list-packages.
>
>     3. I see   matlab-mode        20180928.1526 dependency Major mode for MATLAB(R) dot-m files
>
>     4. I type ? Because I want to understand the dependency but I
>        obtain an error (debug-on-error t) which I attach. 

(I'm only addressing everything up to step 4 below.)

matlab-mode is a dependency of the org-babel-eval-in-repl on MELPA.
So I installed the org-babel-eval-in-repl package, and tried your
recipe. And I got a backtrace on current master.

But I'm getting a different backtrace than you did:

Debugger entered--Lisp error: (file-missing "Opening input file" "No such file or directory" "/home/skangas/.emacs.d/elpa/matlab-mode-20200106.2...")
  insert-file-contents("/home/skangas/.emacs.d/elpa/matlab-mode-20200106.2217/matlab-mode.el")
  lm-commentary("/home/skangas/.emacs.d/elpa/matlab-mode-20200106.2217/matlab-mode.el")
  package--get-description(#s(package-desc :name matlab-mode :version (20200106 2217) :summary "Major mode for MATLAB(R) dot-m files" :reqs nil :kind nil :archive nil :dir "/home/skangas/.emacs.d/elpa/matlab-mode-20200106.2..." :extras ((:keywords "matlab" "programming" "language" "(X)emacs") (:url . "http://sourceforge.net/projects/matlab-emacs/")) :signed nil))
  describe-package-1(#s(package-desc :name matlab-mode :version (20200106 2217) :summary "Major mode for MATLAB(R) dot-m files" :reqs nil :kind nil :archive nil :dir "/home/skangas/.emacs.d/elpa/matlab-mode-20200106.2..." :extras ((:keywords "matlab" "programming" "language" "(X)emacs") (:url . "http://sourceforge.net/projects/matlab-emacs/")) :signed nil))
  describe-package(#s(package-desc :name matlab-mode :version (20200106 2217) :summary "Major mode for MATLAB(R) dot-m files" :reqs nil :kind nil :archive nil :dir "/home/skangas/.emacs.d/elpa/matlab-mode-20200106.2..." :extras ((:keywords "matlab" "programming" "language" "(X)emacs") (:url . "http://sourceforge.net/projects/matlab-emacs/")) :signed nil))
  package-menu-describe-package()
  funcall-interactively(package-menu-describe-package)
  call-interactively(package-menu-describe-package nil nil)
  command-execute(package-menu-describe-package)

Now, the above happens because, when there is no README file,
`package--get-description' expects there to be a (format "%s.el"
package-name) file where we can read the information for installed
packages.

But that's not a convention that the "matlab-mode" package follows for
some reason.  (This requirement/convention does not seem to be
documented in the manual; see (info "(elisp)Multi-file Packages").)

Maybe we need to think a bit about what we can do in these cases.  For
example, there is a matlab.el file that we could have been using
instead.  But the question is if we want to start adding heuristics
here, and if so, which.

Or should we just document the requirement and tell package authors
that they just have to follow it?  Does anyone else have an opinion
here?

   (NOTE: There is a README.org in the matlab-mode git repository[1]
   but it is not included in the MELPA package.  That should probably
   be fixed, presumably by updating the MELPA recipe.)

---

That said, I don't think I understand the below backtrace:

> Debugger entered--Lisp error: (wrong-type-argument char-or-stri3ng-p nil)
>   describe-package-1(#s(package-desc :name matlab-mode :version (20180928 1526) :summary "Major mode for MATLAB(R) dot-m files" :reqs nil :kind nil :archive nil :dir "/home/oub/.emacs.d/elpa/matlab-mode-20180928.1526" :extras ((:keywords "matlab" "programming" "language" "(X)emacs") (:url . "http://sourceforge.net/projects/matlab-emacs/")) :signed nil))
>   describe-package(#s(package-desc :name matlab-mode :version (20180928 1526) :summary "Major mode for MATLAB(R) dot-m files" :reqs nil :kind nil :archive nil :dir "/home/oub/.emacs.d/elpa/matlab-mode-20180928.1526" :extras ((:keywords "matlab" "programming" "language" "(X)emacs") (:url . "http://sourceforge.net/projects/matlab-emacs/")) :signed nil))
>   package-menu-describe-package()
>   funcall-interactively(package-menu-describe-package)
>   call-interactively(package-menu-describe-package nil nil)
>   command-execute(package-menu-describe-package)

Maybe it's just too late for me, and I'm missing something obvious
here... but where does the wrong-type-argument come from?

describe-package-1 has only one argument, and that is the
package-desc.  Cleaning up the package-desc you have, it is basically:

(package-desc-create :name 'matlab-mode
                     :version '(20180928 1526)
                     :summary "Major mode for MATLAB(R) dot-m files"
                     :reqs nil
                     :kind nil
                     :archive nil
                     :dir "/home/oub/.emacs.d/elpa/matlab-mode-20180928.1526"
                     :extras '((:keywords "matlab" "programming" "language" "(X)emacs")
                               (:url . "http://sourceforge.net/projects/matlab-emacs/"))
                     :signed nil)

Which looks very standard to me...

In any case, could you please try this again on current master and see
if you are still seeing the above backtrace, or if you are now seeing
the same as me.  Thanks in advance.

Best regards,
Stefan Kangas

Footnotes:
[1]  https://git.code.sf.net/p/matlab-emacs/src

Previous Next

GNU bug tracking system
Copyright (C) 1999 Darren O. Benham, 1997,2003 nCipher Corporation Ltd, 1994-97 Ian Jackson.