se7entyse7en/pydockenv

Python virtual environment, but backed by Docker!

pydockenv's Issues

Use a single alpine/sockat container for port mapping for each environment

Currently, the container corresponding to an environment is started without mapping any port. Mapping a port would require a restart of the container. To avoid the restart, port mapping is achieved by forwarding the tcp traffic using an alpine/socat container. But currently a container for each port is spun up, this could be improved by using a single alpine/socat container to handle all the ports forwarding.

consider also handling len(cmd) > 1, cmd[1:] could be passed

pydockenv/internal/executor/executor.go

Lines 49 to 54 in dbda5be

	// TODO: consider also handling len(cmd) > 1, cmd[1:] could be passed
	// as arguments to the alias
	alias, aliasExecOptions, err := lookupAlias(cmd[0], contInfo.Config.Labels["aliases"], options)
	if err == nil {
	return ExecuteForContainer(contInfo, alias, aliasExecOptions)
	}

---

This issue was generated by todo based on a TODO comment in dbda5be when #92 was merged. cc @se7entyse7en.

Use a single container for all port mappings instead of

pydockenv/internal/executor/executor.go

Lines 266 to 271 in dbda5be

	// TODO: Use a single container for all port mappings instead of
	// spinning a container for each port
	contName := fmt.Sprintf("%s_port_mapper_%d", cj.Name, p)
	ctxLogger := logger.WithFields(logrus.Fields{
	"port": p,
	"container-name": contName,

---

This issue was generated by todo based on a TODO comment in dbda5be when #92 was merged. cc @se7entyse7en.

Write a proper README

The README should have the following structure:

• Installation
• Why?
  • virtualenv vs pydockenv
• Quickstart
• Examples
  • Basic examples
  • Advanced examples (something like a Flask web app that requires to expose a port)

Support for aliasing

For example aliases:

• python -> pydockenv python

Installation from requirements.txt

Use pydockenv for the project itself

Currently, there's no easy way to use pydockenv for the development of pydockenv itself.

Some ideas:

1. bind the socket into the env container in order to enable the usage of docker from within. See this. The problem is that there's currently a check on the current working environment for running commands relatively to the project root directory.
2. use docker-in-docker. This scenario is much more similar to the real usage of the final user. It will though require the installation, but it may have some problems, see here.

Expose ports to the host

Maybe this can be done by simply setting network to host.

Adds the possibility to use a custom base image and install python on that

cannot install from pypi source tarball (0.2.1)

project cannot be installed from pypi release 0.2.1 since HISTORY.md is used by setup.py but missing in tarball.
Easily replicable by forcing installation from source tarball:

 pip install pydockenv==0.2.1 --no-binary :all:

Autocompletion for commands inside the container

Some "safe" commands should bypass path check

Better error handling

This issue is very generic. Just as an example if an error occurs while an env is being created, eventually created resources (container, network, etc.) are not deleted.

fails with "unsupported shell" when sh is dash

On my system /bin/sh is dash shell.

Test for shell compatibility at the start of pydockenv binary fails since the script is started with /usr/bin/env sh

Add integration test for `create` with toml file

Expand toml file usability by permitting to describe dependencies as unpinned and with other operators (>=, >, etc.)

When a port mapper for a given port is already running on same network don't recreate it

Currently when a port mapper is already running on a given port for an environment, if the same port is passed on another run, then an error is raised. This shouldn't happen, and the creation should be just skipped.

Increase python version compatibility

Replace `envs.json` with container's labels

The meta information of each environment is saved inside <some-path>/envs.json. In case that this file gets deleted or corrupted, it won't be possible to run certain pydockenv commands as many of them require reading it.

This could be improved by removing the need for this file in favor of adding this metadata to the container itself using labels during its creation.

Add tests

Setup for release on pypi

Change the PS1 on activation

Consider using a custom logging handler that wraps click.echo

Travis CI requires pulling docker image first

On Travis CI, if the alpine/socat:latest image is not pulled before running the tests, then the tests fail as it won't find the requested image. The expected result is that the missing image is pulled when requested as it happens locally.

Rename `run` command to `shell` and use `run` for arbitrary commands

Refactor code to make it usable for 3rd party libraries

Currently the code is all in the pydockenv/commands/pydockenv.py file. Better to this after testing (#13 and #8).

Add test that ensures the port mapper works correctly

When listing environments, mark with a `*` the current active

Setup travis for CI

Printing of execution status of the commands only for debugging

Requirements issues

Looking to give this a go, I installed via pypi as directed in the README.

Firstly, on attempted creation I got a massive stack trace which came down to a missing requirement of docker-py. So I installed this with pip install --user docker-py.

Attempting to create a pydockenv again, I get

Creating environment dockenv with python version latest...
Traceback (most recent call last):
  File "<HOME>/.local/lib/python3.7/site-packages/pydockenv/commands/pydockenv.py", line 34, in create
    image = client.images.get(image_name)
AttributeError: 'function' object has no attribute 'get'

During handling of the above exception, another exception occurred:

Traceback (most recent call last):
  File "<HOME>/.local/lib/python3.7/site-packages/pydockenv/commands/pydockenv.py", line 362, in <module>
    cli()
  File "<HOME>/.local/lib/python3.7/site-packages/click/core.py", line 764, in __call__
    return self.main(*args, **kwargs)
  File "<HOME>/.local/lib/python3.7/site-packages/click/core.py", line 717, in main
    rv = self.invoke(ctx)
  File "<HOME>/.local/lib/python3.7/site-packages/click/core.py", line 1137, in invoke
    return _process_result(sub_ctx.command.invoke(sub_ctx))
  File "<HOME>/.local/lib/python3.7/site-packages/click/core.py", line 956, in invoke
    return ctx.invoke(self.callback, **ctx.params)
  File "<HOME>/.local/lib/python3.7/site-packages/click/core.py", line 555, in invoke
    return callback(*args, **kwargs)
  File "<HOME>/.local/lib/python3.7/site-packages/pydockenv/commands/pydockenv.py", line 35, in create
    except docker.errors.ImageNotFound:
AttributeError: module 'docker.errors' has no attribute 'ImageNotFound'

Solved this with pip install --user docker

Then

 Traceback (most recent call last):
  File "<string>", line 1, in <module>
  File "<HOME>/.local/lib/python3.7/site-packages/pydockenv/commands/pydockenv.py", line 13, in <module>
    client = docker.from_env()
  File "<HOME>/.local/lib/python3.7/site-packages/docker/client.py", line 85, in from_env
    timeout=timeout, version=version, **kwargs_from_env(**kwargs)
  File "<HOME>/.local/lib/python3.7/site-packages/docker/client.py", line 40, in __init__
    self.api = APIClient(*args, **kwargs)
  File "<HOME>/.local/lib/python3.7/site-packages/docker/api/client.py", line 128, in __init__
    config_dict=self._general_configs, credstore_env=credstore_env,
TypeError: load_config() got an unexpected keyword argument 'config_dict'
python: can't open file 'create': [Errno 2] No such file or directory

Which version of docker/docker-py is compatible with the project?

For reference, pip freeze | grep dock gives:

docker==3.7.2
docker-py==1.10.6
docker-pycreds==0.4.0
pydockenv==0.2.1

Use isort for automatic handling of imports

Automatically move to project dir on activation

Saving and loading of an environment

Find a way to handle dev deps and prod deps

Use a `pydockenv.toml` to describe an environment

Reference: https://www.python.org/dev/peps/pep-0518/.

Early exit with unsupported shells

Add integration test for `save` and `load` commands

Fix assets on pypi

Consider having subcommands for env management

For example: pydockenv env [create,delete,list].

Create a docker network for each environment

Add a Make target that does everything for release

make all-<part>:

• make publish-check
• git checkout -B release
• make bump-<part>
• create tag
• make publish