Regenerate metadata on update

[maxpowis-bp]

Hi,

I have an openapi-python-client implementation scenario that allows creating python client codebase for a collection of openapi schema's. The implementation will require git repositories to be set up for each of those api client codebases so I can manage their lifecycle independently. The thing is, I can not use the "generate" command as the git repository is dynamically fetched as part of the process, hence, the directory exists at some point and the generate command doesn't allow for a directory to exist upfront. I've managed to have the implementation work in "update" mode by manually instantiating the package directory so the update process doesn't fail. The problem is, I want the pyproject.toml to be part of the generation by that file (and other meta related capabilities) are only created with the "generate" command.

I am actually not sure why the update command doesn't update the metadata as there could be scenerio's where the pyproject.toml.jinja (.gitignore.jinja, ...) templates are updated in upgrades of the openapi-python-client library.

Is there any reason for this or could we actually consider having metadata regenerated as part of the update process?

[dbanty]

The main reason for the existing behavior is that, typically, folks will want to add a bunch of additional metadata to their pyproject.toml, like authors or classifiers. Regenerating would erase anything that's been added (likewise for README).

I think the best bet is probably to implement something like #826. Then we can set a default list of files to ignore, but folks can tweak if it they do want regeneration of certain things.

[dbanty]

update is going away. Instead there will be --output-path and --overwrite to achieve a similar alternative. For more advanced use-cases, we'll pursue #826 combined with --overwrite