Question

I’m learning emacs lisp and I’m trying to script using it. I wrote a script and it works fine but I just think there are a lot of things I make in bash that I can do in emacs lisp instead.

Big deal here: I’m not sure if my start-process works correct

Please suggest / show me the lisp way of scripting on my script (as example) :

#!/usr/bin/emacs --script
(message "Vision synchronization \n")
(let ((default-directory "/home/vision/"))
  (shell-command "git pull;")
  (princ (shell-command-to-string "git fetch upstream;git merge upstream/master;"))
  (princ (start-process "Vision push \n" "git" "git" "push")))

(message "Gentoo-haskell synchronization \n")
(let ((default-directory "/home/gentoo-haskell/"))
  (shell-command "git pull;")
  (princ (shell-command-to-string "git fetch upstream;git merge upstream/master;"))
  (princ (start-process "Gentoo-haskell push \n" "git" "git" "push")))

(message "Nengraphy synchronization \n")
(let ((default-directory "/home/nengraphy/"))
    (princ (start-process "Nengraphy pull \n" "git" "git" "pull")))

(message "Gentoo synchronization \n")
(let ((default-directory "/usr/portage/"))
  (message "Gentoo rsync (New files will be added, deprecated files will be deleted) : \n")
  (princ (shell-command-to-string "rsync --recursive --links --safe-links --perms --times --compress --force --whole-file --delete --timeout=180 --exclude=/.git --exclude=/metadata/cache/ --exclude=/distfiles --exclude=/local --exclude=/packages rsync://209.177.148.226/gentoo-portage/ /usr/portage/"))
  (message "We want to make extra-sure that we don't grab any metadata, since we don't keep metadata for the gentoo.org tree (space reasons)")
  (shell-command "[ -e metadata/cache ] && rm -rf metadata/cache")
  (shell-command "[ -e metadata/md5-cache ] && rm -rf metadata/md5-cache")
  (message "the rsync command wiped our critical .gitignore file, so recreate it.")
  (shell-command "echo \"distfiles/*\" > /usr/portage/.gitignore")
  (shell-command "echo \"packages/*\" >> /usr/portage/.gitignore")
  (message "profile formats fix")
  (shell-command "echo \"profile-formats = portage-1\" >> /usr/portage/metadata/layout.conf")
  (message "\"git add .\" will record all the changes to local files the git repo. So there must be no stray files.")
  (shell-command "if [ ! -d profiles/package.mask ]
  then
    mv profiles/package.mask profiles/package.mask.bak || exit 4
    install -d profiles/package.mask || exit 4
    mv profiles/package.mask.bak profiles/package.mask/gentoo || exit 4
  fi")
  (princ (shell-command-to-string "git add ."))
  (message "create a commit")
  (shell-command "git commit -a -m \"gentoo updates `date` update\"")
  (message "push these changes up.")
  (princ (shell-command-to-string "git push origin master")))

(message "Gentoo verification \n")
(princ (shell-command-to-string "emerge --sync;"))

(message "Layman synchronization \n")
(princ (shell-command-to-string "layman -S;"))

thank you!

Answer 1

This is really just a bash script using some awkward elisp wrapping. There is no advantage to using elisp this way. Until you start using elisp to actually compute something, you should probably stick with bash.

You could use emacs to script opening text files and modifying the text, like what you might do with sed or awk. That would make a little more sense. At this point, it would be faster for me to do this with elisp than to look up the awk syntax.

You do have to start somewhere, but what you are doing is a very awkward fit for elisp. Better to learn by working with the strengths of a language than by forcing it to do something it is not really made for.

Also, start-process does not make sense in a script. It starts a process within emacs to allow for ongoing communication, which you do not want or need in a script.