start-session

#!/bin/sh

### Session workflow
#
#   NB: you need to run all the mentioned scripts from the repo root.
#
# * Checkout starting point. For a fresh session the `main` branch is a good choice.
# * Make sure that the git tree is clean, as this script will be committing your
#   changes as you make them.
# * Start server with `./start-server`
# * Run this script with a session name as an argument.
#   It must be valid as a part of git branch name as the script will prefix it with
#   `session/` and checkout this branch.
# * If the session with such name exists it will be resumed.
# * Make changes in `src/session.nim` and save file to compile and send to the server.
# * After every successful compilation there will be a commit.
# * Ctrl-C this script to stop and switch back to the starting point.
# * The system is ready for the next session!
#
###

if [ -z $1 ]; then
	echo "Please provide session name."
	exit 1
fi

if ! git diff --quiet HEAD; then
	echo "Tree must be clean."
	exit 1
fi

git checkout "session/$1" || git checkout -b "session/$1" || exit 1

rm target/session.so.*

echo "All set, make sure server is running and edit src/session.nim to make some noises."

# dlopen tries really hard to reuse shared objects. The simplest way I've found
# to workaround it on macOS is to build the library with a different name each
# time. Just "double-buffer" copying as clive does didn't work, I assume because
# not only the filename/inode is checked but LC_ID_DYLIB baked into the file
# too. We also have to randomize the naming and remove unloaded libs to
# make it work.

fswatch --event=Created --event=Updated -e flycheck -o src | while read num; do
	rm target/session.so.*
	nim c --app:lib -o:target/session.so.$RANDOM src/session.nim && git commit -am "$(date)"
done

git commit -am "$(date)"
git checkout -