Lute v3 alpha release up on test.pypi.org

Lute v3 is getting pretty close to ready:

• All tests ported: check
• All features implemented: check (I think)
• All obvious codo TODOs addressed: check
• Packaged and up on testpypi: check
• Dogfooded and user feedback implemented: not yet
• Documentation updates (porting, wiki updates): not yet
• etc …

So I guess it’s time to start dogfooding and to get some user notes.

Time for user feedback?

Lute v3 is a full rewrite in Python, and is (initially) available as alpha software at https://test.pypi.org/project/lute3/.

test.pypy.org is a testing ground before I publish the “for-realsies” lute3 application to the official PyPi. I may have messed up some packaging things, so keeping it on test.pypi.org helps keep the release clean.

It’s called lute3, and not just lute, because there’s already a (junk) package called lute up on PyPi.

Pre-reqs

• lute3 requires Python 3.11, as far as I know. The package specifies 3.11+ as a requirement … perhaps that can be relaxed, I’m honestly not sure.
• You’ll need pip3 to install it, again as far as I know.

Installation using pip3

While it’s not a necessity, I assume you’ll install into some kind of virtual/sandboxed environment.

E.g., using python3 venv, in some empty directory:

python3 -m venv .venv
source .venv/bin/activate

If you don’t want to use a virtual environment, you don’t have to.

Then, get the stuff. test.pypi.org doesn’t have all of the packages lutev3 needs, so you need to install it like this.

# In your virtual environment, or not :-)
pip3 install --upgrade --index-url https://test.pypi.org/simple/ --extra-index-url https://pypi.org/simple/ lute3

Running

Assuming you’ve set up a virtual environment and installed lute3:

source .venv/bin/activate
python -m lute.main

This will start a python process and print a short message:

python -m lute.main
  
  Starting Lute:
  
  data path: /Users/jeff/Library/Application Support/Lute3
  database: /Users/jeff/Library/Application Support/Lute3/lute.db
  
  Running at:
  
  http://localhost:5000
  
  When you're finished reading, stop this process
  with Ctrl-C or your system equivalent.

Open your web browser to http://localhost:5000 and do your thing … then when done, go back to that terminal/command line running the python -m lute.main, and hit Ctrl-C to stop it. Then deactivate to exit/deactivate the virtual environment if you’re using one.

DON’T USE THIS FOR YOUR REAL DATA JUST YET

If you’re feeling sassy, sure, copy your data files from your current lute location to the data path shown in the “Starting Lute” message (or in the “software info” link on the main screen).

BUT don’t move the data. Two reasons:

1) Lute v3 makes a few changes to the db schema that v2 won’t support. If you move your db to v3, there’s no moving back to v2 – well, not trivially.

2) Lute v3 is pretty damn good, but it’s still alpha. I haven’t dogfooded¹ it yet on my own data. I’ve used it on test data, and the algorithms and tests are the same as were used in Lute v1 and v2, so it should be good. But you never know! And I’m going to repackage it eventually anyway for the real pypi release.

What’s different?

Honestly, right now, not much, from the user’s perspective:

• No more .env file. Backup settings are managed through the UI
• Data is – or should be – stored in the appropriate spot, as determined by your operating system. Lute v3 uses a Python library called PlatformDirs to determine where to store data. The approach should ensure a relatively smooth installation.
• Configuration for backup settings and custom styles are managed through a “settings” link, as opposed to external files.
• ummmm … the Lute logo now has three little slashes, to mean “3” …

From the back-end perspective, of course, everything is completely different. This is a full rewrite (again!), and IMO the project has taken a big step in the right direction.

No Docker just yet

I’ve been scrambling to get the TestPyPi install right, as I feel that’s the right first step for packaging. I haven’t made a Dockerfile or Docker image.

Docker work should be pretty straightforward:

• find a base image with Python 3.11
• the lute image build process should:
  • copy the lute/config/config.yml.docker file to lute/config/config.yml
  • specify some container folder as the config.yml DATAPATH
  • specify some container folder as a backup folder
  • blah blah entry point yadda yadda
• a docker-compose.yml file should mount host folders to the datapath and backup folders

Final state

Well, who knows, we’ll get there when we get there. I expect some kind of thing like:

• a desktop installer (maybe using pyinstall or whatever it is, so really non-techie users double-click to run it.
• PyPi package
• Docker for super-keeners
• Source code installs.

Cheers!