Bring in gxformat2 Linting Improvements (add some agent assistance files)#1633
Bring in gxformat2 Linting Improvements (add some agent assistance files)#1633jmchilton wants to merge 3 commits intogalaxyproject:masterfrom
Conversation
| @@ -0,0 +1,21 @@ | |||
| Walk me through readying a planemo release. Run each step, check for problems, and pause if anything looks wrong. | |||
|
|
|||
| ## Steps | |||
There was a problem hiding this comment.
Can we point to https://planemo.readthedocs.io/en/latest/developing.html or merge with ?
There was a problem hiding this comment.
This was generated from that (or from gxformat2's version of that since that is where I got the command) but I can link the doc and request they stay in sync. Or would you rather this just say "read the doc, confirm the preconditions, and then run release method specified in the doc". I guess the answer is a bit of both huh... there should be a Makefile target that ensures the tree is clean, etc...?
Ugh... I will remove this command too and look at overhauling it in the context of gxformat2 first.
There was a problem hiding this comment.
i was hoping for read the doc, confirm the preconditions, and then run release method specified in the doc". ... one source of truth would be good.
There was a problem hiding this comment.
I obviously agree about the source of truth.
I want to work on docs/developing.rst and the /ready-release command a bit. I think the developing doc is
very out of date - we don't use Travis anymore, we probably don't need to check the docs before the
release, we use trusted publishing, etc... Can you give me a plan that includes a new developing.rst,
adding a reference to the docs in /ready-release, a new Makefile target that does a pre-release checklist
(have history entries, .venv exists, git is clean, "make clean && make lint" passes. developing.rst should
not mention the agent instructions and the agent command should try to minimize how much information is
stored in the agent file and try to defer to the documentation.
| @@ -0,0 +1,51 @@ | |||
| # CLAUDE.md | |||
There was a problem hiding this comment.
Does claude get any of this wrong ? On the surface that seems like stuff claude does without telling it to, but it consumes tokens on every session.
There was a problem hiding this comment.
No you're right. This is the default init as of last Thursday - I read they recently overhauled init and it was better now. But double checking on that just now apparently it has to be enabled and isn't on by default. https://medium.com/@AdithyaGiridharan/claude-codes-init-command-is-getting-a-brain-transplant-e0c389ec0c6b
I will drop this from the PR.
mvdbeek
left a comment
mvdbeek
left a comment
There was a problem hiding this comment.
This is awesome, minus the claude.md. If we add anything it should be very short (use makefile when appropriate) ? xref https://arxiv.org/html/2602.11988v1
Bump gxformat2>=0.25.0; switch workflow_lint to lint_*_path and
lint_best_practices_{ga,format2}; delete the duplicated planemo
_lint_best_practices body that was migrated upstream. Wire the new
lint_pydantic_validation as a schema_validation lint step. Coerce
gxformat2 Linter classes to their names in WorkflowLintContext so
galaxy's LintMessage doesn't render class reprs.
Update test assertions for the new message text (drops "with ID",
uses native step ids directly). Clean up three legacy test fixtures
that strict pydantic validation now rejects: null input key in
wf14-unlinted-best-practices.yml and format2-style top-level inputs/
outputs on two native basic_native fixtures.
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
jmchilton
force-pushed
the
linting_update
branch
from
bb1b705 to
1ae7a45
Compare
- Handle list-valued input format and extra scalar type names (integer, text, long, double) from updated inputs_normalized output. - Drop gxformat2.interface usage; galaxy_interface is now unused by python_to_workflow. Removes DummyImporterGalaxyInterface and BioBlendImporterGalaxyInterface wiring; import_workflow no longer takes admin_gi. Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
|
test_autoupdate_gxformat2_workflow times out, i wonder if that could be related ? |
2 participants
Use gxformat2 0.25 lint interface; drop duplicated best-practices
Bump gxformat2>=0.25.0; switch workflow_lint to lint_*path and
lint_best_practices{ga,format2}; delete the duplicated planemo
_lint_best_practices body that was migrated upstream. Wire the new
lint_pydantic_validation as a schema_validation lint step. Coerce
gxformat2 Linter classes to their names in WorkflowLintContext so
galaxy's LintMessage doesn't render class reprs.
Update test assertions for the new message text (drops "with ID",
uses native step ids directly). Clean up three legacy test fixtures
that strict pydantic validation now rejects: null input key in
wf14-unlinted-best-practices.yml and format2-style top-level inputs/
outputs on two native basic_native fixtures.