Linter improvements by h-vetinari · Pull Request #2473 · conda-forge/conda-smithy

h-vetinari · 2026-02-26T06:43:54Z

What started out as just wanting to rip out the logic for MACOSX_DEPLOYMENT_TARGET became a deeper and deeper rabbit hole as various obvious follow-up clean-ups were blocked by the state of the code.

This PR:

changes the logic regarding OSX pins (c_stdlib_version / MACOSX_SDK_VERSION) to simply ignore MACOSX_DEPLOYMENT_TARGET; this showed that the merge logic contained some subtle bugs where our tests were misconfigured after reflect upcoming macos 11.0 baseline #2461, but didn't trigger (I don't expect this to affect many recipes, but strictly speaking we were missing some lints)
break out osx-specific lints from lint_stdlib to a separate lint that's conda-forge-specific
in that new check, also lint against presence of MACOSX_DEPLOYMENT_TARGET
separately, also lint the content of conda_build_config.yaml in v1 recipes (e.g. the state before Rebuild for aws-crt-cpp 0.37.3 aws-sdk-cpp-feedstock#993 didn't actually trigger a linter warning despite the outdated pins)
add lint if both conda_build_config.yaml as well as variants.yaml are in the recipe folder
consolidate and clean up a bunch of redundant logic in the main linter driver.
skip a linter test that fails on windows

All the commits have been individually tested and I highly recommend to review them one-by-one (consider each as a mini-PR; the overall diff is too messy).

now that it no longer mixes hints and lints

move it to conda-forge-specific lint

h-vetinari · 2026-02-26T07:23:08Z

Any idea what's going on with coveralls here? 🤔

Run coverallsapp/github-action@master
Using lcov file: ./coverage.lcov
Error: Bad response: 530 error code: 1016

A restart didn't fix it...

CC @beckermr @jaimergp

xhochy

Content-wise, this looks good; two minor nits that don't need to be fixed.

xhochy · 2026-02-26T08:14:35Z

conda_smithy/lint_recipe.py

        )

-    # 8. check for obsolete os_version
+    # 7. check for obsolete os_version


Unrelated nit: We're a bit inconsistent in the comments with "1." vs "1:" ;)

Personally, I'd be more inclined to just remove the numbering. It serves no purpose, and makes it harder to move lints (if it ever becomes desirable to group or otherwise reorder checks)

xhochy · 2026-02-26T08:15:48Z

conda_smithy/lint_recipe.py

-        conda_build_config_keys = set()
-        forge_yaml = {}
+    # Interlude: load feedstock and recipe config
+    feedstock_config_keys = _get_feedstock_config(recipe_dir)


What's the reason you only load this here and not at the very top of the function?

This PR was already ballooning in size, so I tried to keep changes local where possible. Further cleanups are of course possible.

h-vetinari

Thanks for the review!

h-vetinari · 2026-02-26T08:34:30Z

conda_smithy/lint_recipe.py

        )

-    # 8. check for obsolete os_version
+    # 7. check for obsolete os_version


Personally, I'd be more inclined to just remove the numbering. It serves no purpose, and makes it harder to move lints (if it ever becomes desirable to group or otherwise reorder checks)

h-vetinari · 2026-02-26T08:36:03Z

conda_smithy/lint_recipe.py

-        conda_build_config_keys = set()
-        forge_yaml = {}
+    # Interlude: load feedstock and recipe config
+    feedstock_config_keys = _get_feedstock_config(recipe_dir)


This PR was already ballooning in size, so I tried to keep changes local where possible. Further cleanups are of course possible.

h-vetinari · 2026-02-27T02:27:56Z

Ah, the error finally has some more context now.

Error: Bad response: 503 <!DOCTYPE html>
<html lang="en">
<head>
  ...
</head>
<body>
  <div class="container">
    <h1>Please Hold</h1>
    <p>We're performing some unscheduled maintenance. Please check our status page for updates: <a href="https://status.coveralls.io/">Status Page</a> </p>
  </div>
</body>
</html>

"unscheduled maintenance" does not sound good. In fact, https://status.coveralls.io/ says

after two full days of working directly with our hosting provider’s account team, we still do not have an ETA for service restoration.

So I guess we can ignore this here, though we might want to break out the coverage stuff into a separate check, so that at least it becomes visible at first glance whether the tests passed or not.

h-vetinari · 2026-02-27T03:25:01Z

Thanks for the review @xhochy!

h-vetinari added 12 commits February 26, 2026 17:08

stop considering MACOSX_DEPLOYMENT_TARGET

dfedb01

move lint_stdlib to correct section

9d46680

now that it no longer mixes hints and lints

avoid else-branches that just set default values for reading CBC

67f89b4

clarify naming of config files in lint_recipe.py

5c32a87

reuse find_local_config_file for _get_forge_yaml

1f0c41b

also lint conda_build_config.yaml for v1 recipes

f596a98

unify ways linter gets feedstock config

71066d7

unify ways linter gets recipe config

9e3919f

add lint for recipes containing several recipe config files

ef3e912

break out lints for osx pins into separate function

d9f2a02

move it to conda-forge-specific lint

add lint for remaining occurrences of MACOSX_DEPLOYMENT_TARGET

8647c27

skip a CRLF-sensitive test on windows

457b6c0

h-vetinari requested a review from a team as a code owner February 26, 2026 06:43

xhochy reviewed Feb 26, 2026

View reviewed changes

h-vetinari commented Feb 26, 2026

View reviewed changes

fix docstring of _get_recipe_config_keys

8ab91bb

h-vetinari added 2 commits February 27, 2026 13:31

add reference to PR in news item

d8dd502

don't fail on coveralls, for now

ea1897f

h-vetinari force-pushed the no_dt branch 2 times, most recently from fb8840c to ea1897f Compare February 27, 2026 03:13

switch to coveralls main branch

a90931d

h-vetinari merged commit 5dd5767 into conda-forge:main Feb 27, 2026
2 checks passed

h-vetinari deleted the no_dt branch February 27, 2026 03:25

This was referenced Feb 27, 2026

Re-enable coveralls #2474

Open

RFC: rename main linter functions #2475

Open

Status Report (2026-02-27) Quansight-Labs/conda-ecosystem-sta-mgmt#118

Open

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Uh oh!

Linter improvements#2473

Linter improvements#2473
h-vetinari merged 16 commits intoconda-forge:mainfrom
h-vetinari:no_dt

h-vetinari commented Feb 26, 2026 •

edited

Loading

Uh oh!

h-vetinari commented Feb 26, 2026

Uh oh!

xhochy left a comment

Uh oh!

xhochy Feb 26, 2026

Uh oh!

h-vetinari Feb 26, 2026

Uh oh!

xhochy Feb 26, 2026

Uh oh!

h-vetinari Feb 26, 2026

Uh oh!

h-vetinari left a comment

Uh oh!

h-vetinari Feb 26, 2026

Uh oh!

h-vetinari Feb 26, 2026

Uh oh!

h-vetinari commented Feb 27, 2026

Uh oh!

Uh oh!

h-vetinari commented Feb 27, 2026

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

2 participants

Uh oh!

Conversation

h-vetinari commented Feb 26, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

h-vetinari commented Feb 26, 2026

Uh oh!

xhochy left a comment

Choose a reason for hiding this comment

Uh oh!

xhochy Feb 26, 2026

Choose a reason for hiding this comment

Uh oh!

h-vetinari Feb 26, 2026

Choose a reason for hiding this comment

Uh oh!

xhochy Feb 26, 2026

Choose a reason for hiding this comment

Uh oh!

h-vetinari Feb 26, 2026

Choose a reason for hiding this comment

Uh oh!

h-vetinari left a comment

Choose a reason for hiding this comment

Uh oh!

h-vetinari Feb 26, 2026

Choose a reason for hiding this comment

Uh oh!

h-vetinari Feb 26, 2026

Choose a reason for hiding this comment

Uh oh!

h-vetinari commented Feb 27, 2026

Uh oh!

Uh oh!

h-vetinari commented Feb 27, 2026

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

2 participants

h-vetinari commented Feb 26, 2026 •

edited

Loading