Re: [O] Advice needed: Cant find a decent way to autosync my work TODO org file between laptop and work pc

[Richard Lawrence]

Hi Xebar,

Xebar Saram <address@hidden> writes:

> i mainly used git over the last 6 months but that forces me to
> pull/commit/push manually each time i add something to either machine and
> that is really annoying. plus i get merge conflicts all the time
>
> I have tried dropbox at the past but again this causes conflicts,
> isntreliable to me (on my
> linux laptop it doesn't always resume sync after sleep)

If neither git nor Dropbox works for you, I'm not sure this will be
helpful, but...

I personally use git for this.  To keep the effort of syncing and
merging low, I have a cron job set up to automatically commit in certain
repositories containing my Org files every hour.  It's just a
quick-and-dirty hack, but here it is:

#+BEGIN_SRC bash
#!/bin/bash
# Add org file changes to the repository
ROOT=$HOME
REPOS="org Documents/philosophy/dissertation Documents/philosophy/teaching 
Documents/philosophy/reading src/emacs"

commit_and_push()
{
  for REPO in $REPOS
  do
      echo "Repository: $ROOT/$REPO"
      cd $ROOT/$REPO

      if [ ! -d .git ]; then
          echo "Not a git repository; skipping"
          continue
      fi
      if [ ! -r autocommits ]; then
          echo "No autocommits file found; skipping"
          continue
      fi

      # Remove deleted files
      git ls-files --deleted -z | xargs -0 git rm >/dev/null 2>&1

      # Add files on whitelist to commit for current branch
      while read FILE
      do
          git add $FILE
      done < autocommits

      git commit -m "$(date) from $(hostname) by autocommit.sh"
      STATUS=$?
      if [ $STATUS != 0 ]; then
          echo "git commit failed with exit status $STATUS"
      fi

      # Push the current branch
      # requires branch.<name>.remote to be specified in git config
      # and ideally push.default = tracking 
      git push --porcelain
      STATUS=$?
      if [ $STATUS != 0 ]; then
          echo "git push failed with exit status $STATUS"
      fi

  done
}

pull()
{
  for REPO in $REPOS
  do
      echo "Repository: $REPO"
      cd $ROOT/$REPO

      git status | grep 'modified:'
      if [ $? == 0 ]; then
          echo "modified files present; fetching, but not pulling."
          CMD="fetch"
      else
          CMD="pull"
      fi

      git $CMD
      STATUS=$?
      if [ $STATUS != 0 ]; then
          echo "git $CMD failed with exit status $STATUS"
      fi
  done
}

case "$1" in
  commit)
      commit_and_push
      ;;
  pull)
      pull
      ;;
  *)
      echo "Usage: $0 {commit|pull}"
      exit 1
      ;;
esac

exit 0
#+END_SRC

This script requires a file called "autocommits" in each repository in
$REPOS which is a whitelist of files to autocommit changes in.  Most of
my Org files are on such a list.

I run this from Cron like:
#+BEGIN_SRC cron
0 * * * * ~/bin/autocommits.sh commit >/dev/null 2>&1 
#+END_SRC

To make sure my changes are saved before this script runs, I have in my .emacs:
#+BEGIN_SRC elisp
(run-at-time "00:59" 3600 'org-save-all-org-buffers) ; cron commits on the hour
(add-hook 'before-save-hook 'org-update-all-dblocks)
#+END_SRC

And finally, to pull the autocommitted changes on a new machine:
$ autocommits.sh pull

So you still have to manually pull and fix any merge conflicts, but I
find that committing often generally keeps the work involved here
minimal (because I can't usually work on two machines at once!).

Hope that helps!

Best,
Richard