Installation strategy without new releases?

[cluhmann]

I am writing beangulp-based importers and following the examples mentioned in the README. However, the installation suggested there (i.e., pip install beangulp) pulls the most recent release that is now a year old (v0.2.0). The beancount installation doc found here has no mention of beangulp. Are new beangulp releases planned? Or is another installation method recommended?

[MickMorley]

Glad you posted this. I was on a wild goose chase wondering why my amount = csvbase.Amount(name="Amount", negate=True) code was not working. I found out that the version in pip does not yet have this functionality.

I found out the same on this page https://pypi.org/project/beangulp/#history.

[MickMorley]

I found that fava installs beangulp 0.2.0.

You can run pip install git+https://github.com/beancount/beangulp#egg=beangulp and it will uninstall version 0.2.0 and install the current version from the repository (as of today: 0.3.0.dev0)

[cluhmann]

This approach might work, but then you are vulnerable to whatever is random commits happen to sit on main (at the point you install from github). Maybe the maintainers take great care to ensure that main is always "release-ready" (e.g., trunk-based development), but if that's the case then the installation instructions should be changed (and it's unclear why there were releases in the past). Otherwise, your local installation may pull random half-finished or broken code. That's why I was asking for clarification on what exactly people are expected to do here.

[trim21]

you can pin the sha pip install https://github.com/beancount/beangulp/archive/1f315a62afc5160bb1d62759451f3610186e293f.zip

or

[project]
dependencies = [
  "beangulp @ https://github.com/beancount/beangulp/archive/1f315a62afc5160bb1d62759451f3610186e293f.zip",
]