july3 is Make with Python syntax. See this july3file.py example.

@Rule(f'test-build/{env.project_name}', dependencies=['files/nginx-site.mako', 'test-build'])
def nginx_site_file(rule):
    with open(rule.name, 'w') as f:
        f.write(Template(filename=rule.dependencies[0]).render(**env))


@Rule('test-build')
def build_dir(rule):
    os.makedirs(rule.name)

You can build rule like this:

july3 nginx_site_file

july3file.py and july3 command works like Makefile and make command.

july3 is developed to substitute Fabric. It's a good deployment automation tool, but the writer of fabric is trying to separate it as two parts, ssh streamline and task managemant. ssh streamline part will be fabric 2.0, and task management became invoke. I agree with the direction of changes, but I don't like invoke. It doesn't provide reliable dependency management, so each task should take care of idempotence by itself.

However, Make provides good way to define and calcuate dependencies. Actually you can write decent build description with Make. It has better design than modern automation tools like Maven, Gradle, Ansible, Chef and etc. You can define rules and dependencies in declarative way but write commands in programmtic way. So you don't need writing some kind of plugins to define new type of rules.

On of the weak points of Make is unfriendly syntax. Of course the rule-dependencies-command structure is easy and clear, but the symbols like $< or $@ are hard to remember, and shell script syntax is not familiar. I always google how to use if statement in bash. Make only provides timestamp of file based dependencies. It works for most cases, but a few cases need other way of detect dependency, like service is running or package installed.

Therefore, I wrote july3 as more friendly Make, not as yet another Ant, Maven, Gradle, Fabric, Ansible or some other modern build tools.

july3 only support Python 3.x.

pip install july3

Because july3 is started as deployment automation tool and it's syntax is based on Python, it provides some contrib packages related to deployment of python tools.

• python, pip, virtualenv
• nginx, uwsgi
• aiohttp
• celery
• postgres
• ubuntu package