switch to dep groups (PEP 735) by mathematicalmichael · Pull Request #145 · pyOpenSci/pyos-package-template

mathematicalmichael · 2026-01-12T23:22:11Z

I think main.yml should be more explicit about python -m pip install hatch as 1.16+ is definitely required. but I left it be for now.

I ran with uv run --with hatch --with twine hatch... in order to test this, but ultimately want to open the PR to see if CI/CD will pass as well.

it bothers me that twine needed to be added to main.yml but 3bef2e1 didn't address it like I expected. somehow hatch is not properly handling the dep group for itself?

If there's one "hack" "in this PR, that feel like it. perhaps twine belongs elsewhere.

fixes #143

…2 updates Bumps the actions group with 2 updates in the / directory: [actions/checkout](https://github.com/actions/checkout) and [actions/setup-python](https://github.com/actions/setup-python). Updates `actions/checkout` from 5.0.0 to 6.0.0 - [Release notes](https://github.com/actions/checkout/releases) - [Changelog](https://github.com/actions/checkout/blob/main/CHANGELOG.md) - [Commits](actions/checkout@08c6903...1af3b93) Updates `actions/setup-python` from 6.0.0 to 6.1.0 - [Release notes](https://github.com/actions/setup-python/releases) - [Commits](actions/setup-python@e797f83...83679a8) --- updated-dependencies: - dependency-name: actions/checkout dependency-version: 6.0.0 dependency-type: direct:production update-type: version-update:semver-major dependency-group: actions - dependency-name: actions/setup-python dependency-version: 6.1.0 dependency-type: direct:production update-type: version-update:semver-minor dependency-group: actions ... Signed-off-by: dependabot[bot] <support@github.com>

This reverts commit 3bef2e1.

mathematicalmichael · 2026-01-12T23:42:30Z

okay so since tests do pass, I tried to address the "hack" with twine with this:

diff --git a/template/pyproject.toml.jinja b/template/pyproject.toml.jinja
index 0e4e8aa..6d2b60c 100644
--- a/template/pyproject.toml.jinja
+++ b/template/pyproject.toml.jinja
@@ -75,6 +75,7 @@ Download = "https://pypi.org/project/{{ project_slug }}/#files"
 dev = [
     "hatch",
     "pre-commit",
+    "twine",
     "{{ package_name }}[
         {%- if documentation!="" %}docs,{% endif -%}
         {%- if use_test %}tests,{% endif -%}
@@ -138,6 +139,7 @@ types = [
 dev = [
     "hatch",
     "pre-commit",
+    "twine",
 ]
 
 {%- if documentation != "no" %}

and can report that this did not work. like... where else would this go?

I also additionally tried python -m twine check dist/* in place of twine check dist/* in combination.

still:

/bin/sh: twine: command not found

really feels like a broader issue.

what did work was:

# Explicit dependencies are required as a workaround for dependency-groups
# not being installed in builder environments (see https://github.com/pypa/hatch/issues/2152)
# This ensures twine is available when scripts run
dependencies = [
    "twine",
]

mathematicalmichael · 2026-01-12T23:55:47Z

TBH I don't actually have the mental capacity at this moment to decide if 77e0622 is actually the right thing to do (or if it's better to revert it), I'll let someone else make that call.

my understanding of the acceptance criterion was dependencies = [], so the last commit feels like a hack as well, but the only way I got that it work w/o it was installing twine globally, which sort of defeats the purpose anyway (hatch should be all that's needed).

there may be a better solution here.
I'm allowing edits from maintainers bc I'm not sure I'll have the capacity to test further than this.

cc @sneakers-the-rat do you have an opinion on this matter?

…/dep-group

lwasser · 2026-01-16T00:35:52Z

template/pyproject.toml.jinja

 {%- endif %}
 Download = "https://pypi.org/project/{{ project_slug }}/#files"

+{%- if not use_hatch_envs %}


@mathematicalmichael optional-dependencies are for optional features in a package. so they shouldn't be correlated with development-groups which are for development dependencies. I suggest we still include them/remove the conditional here and just add a comment about how they are used.

Suggested change

{%- if not use_hatch_envs %}

i'm trying to find the end of this condition - but we may just need to move it down.

thanks!! yes i think project.optional-dependencies can just be empty in our pyproject.toml template with a comment that the user can use that for optional features if they have them. but it shouldn't correlate to hatch environments because it's just a core pyproject.toml feature and development groups just allow us to separate feature dependencies from development.

Please let me know if i'm missing your intention here!

lwasser · 2026-01-16T00:36:31Z

template/pyproject.toml.jinja

-# The groups below should be in the [development-groups] table
-# They are here now because hatch hasn't released support for them but plans to
-# in Mid November 2025.
+# When not using hatch environments, we keep optional-dependencies for backward compatibility


Suggested change

# When not using hatch environments, we keep optional-dependencies for backward compatibility

# Optional-dependency table can be used to features that support optional functionality in your package. Remove this section if you don't have optional features in your package!

lwasser · 2026-01-16T00:39:13Z

template/pyproject.toml.jinja

-# They are here now because hatch hasn't released support for them but plans to
-# in Mid November 2025.
+# When not using hatch environments, we keep optional-dependencies for backward compatibility
 dev = [


I think this entire section can get moved to development groups but i may be wrong! it's hard to wrap my head around a jinja template in a pr!! but dev definitely belongs in development-groups not optional-dependencies

yeah they are really annoying to review w/o indents, aren't they?

there's already an equivalent dep group for dev on 137 - so yeah perhaps this can get yeeted / was just left over.

lwasser · 2026-01-16T00:43:26Z

template/pyproject.toml.jinja

+# Explicit dependencies are required as a workaround for dependency-groups
+# not being installed in builder environments (see https://github.com/pypa/hatch/issues/2152)
+# This ensures twine is available when scripts run
+dependencies = [
+    "twine",
+]


I need to test this again - the pr to fix this was merged, but when I tested this last, it worked as long as builder was set to true in the environment definition here. Let me do a sanity check on it - @mathematicalmichael, did you run through the template and try it with Hatch 1.16.2 without adding twine explicitly? I'm sorry if I missed a discussion of this, but I thought it worked, but you just had to make it a builder envt. it's possible i just had the newest commit/fix installed and didn't realize it but i thought that was the patch for now - builder=true.

yup. in commit history i tried to get rid of twine explicitly a few times and it kept failing. the only other workaround was to pip install it outside (when installing hatch) - equally bad.

i only pushed up stuff that passed locally to CI. everything else failed.

lwasser

@mathematicalmichael thank you so much for this pr. I will test this - but likely not until next week after the holiday monday. I left a few comments but i want to redo my tests because i've installed various versions and commits where hatch was updated so i want to do a sanity check on what works and what doesn't first!

mathematicalmichael · 2026-01-16T00:46:18Z

template/pyproject.toml.jinja

+# Explicit dependencies are required as a workaround for dependency-groups
+# not being installed in builder environments (see https://github.com/pypa/hatch/issues/2152)
+# This ensures twine is available when scripts run
+dependencies = [


this irks me

lwasser · 2026-01-16T00:49:24Z

I think main.yml should be more explicit about python -m pip install hatch as 1.16+ is definitely required. but I left it be for now.

I agree. I was actually thinking about leaving that comment! At a minimum we want 1.16.2 where dependency group support was added. ✨

dependabot bot and others added 7 commits December 1, 2025 09:57

switch to dep groups (PEP 735)

47794f9

builder=true (for now)

8d13d80

hatch run build:check to scope tests properly

66f4bf1

invoke twine as a module (dont assume it is in PATH)

3bef2e1

Revert "invoke twine as a module (dont assume it is in PATH)"

4c624bb

This reverts commit 3bef2e1.

guess we need twine?

7e546ea

mathematicalmichael marked this pull request as ready for review January 12, 2026 23:29

declare twine as a dep to avoid manual installation

77e0622

mathematicalmichael added 2 commits January 12, 2026 16:58

fix reference to github username

93b4c3f

Merge branch 'dependabot/github_actions/actions-78e82347d6' into feat…

42500e9

…/dep-group

mathematicalmichael mentioned this pull request Jan 13, 2026

dependencies: (deps): bump the actions group across 1 directory with 2 updates #142

Open

sneakers-the-rat mentioned this pull request Jan 15, 2026

switch hatch to dependency groups, split deps to data file #144

Open

lwasser reviewed Jan 16, 2026

View reviewed changes

mathematicalmichael commented Jan 16, 2026

View reviewed changes

	# When not using hatch environments, we keep optional-dependencies for backward compatibility
	# Optional-dependency table can be used to features that support optional functionality in your package. Remove this section if you don't have optional features in your package!

Conversation

mathematicalmichael commented Jan 12, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

mathematicalmichael commented Jan 12, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

mathematicalmichael commented Jan 12, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

mathematicalmichael Jan 16, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

lwasser left a comment

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

lwasser commented Jan 16, 2026

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

2 participants

mathematicalmichael commented Jan 12, 2026 •

edited

Loading

mathematicalmichael commented Jan 12, 2026 •

edited

Loading

mathematicalmichael commented Jan 12, 2026 •

edited

Loading

mathematicalmichael Jan 16, 2026 •

edited

Loading