[build] Use MODULE.bazel for all dependencies

[jwnimmer-tri]

Towards #20731.

The WORKSPACE.bzlmod is no longer load-bearing, so as of this commit it should be possible to consume Drake as module external of other downstream projects (instead of as a repository external).

See RobotLocomotion/drake-external-examples#348 for how to use this downstream.

---

This change is 

[xuchenhan-tri]

+@xuchenhan-tri for feature review.

[jwnimmer-tri]

Reviewed 2 of 2 files at r5, 3 of 18 files at r7, 18 of 18 files at r8, all commit messages.
Reviewable status: LGTM missing from assignee xuchenhan-tri(platform), needs at least two assigned reviewers

---

tools/install/install.bzl line 534 at r8 (raw file):

        content = "\n".join([
            "#!/bin/bash",
            "{} --actions '{}' \"$@\"".format(

FYI This is required because of how our CMake install is coming from a different MODULE now. We can't hard-code the path anymore.

[xuchenhan-tri]

Timed out for the day. I'm slowly working through this. It is taking longer than I expected, and I'll have to come back to this tomorrow.

Reviewed 2 of 10 files at r2, 1 of 7 files at r3, 1 of 2 files at r4, 2 of 2 files at r5, 1 of 18 files at r7, 12 of 18 files at r8, 2 of 4 files at r9, all commit messages.
Reviewable status: 2 unresolved discussions, LGTM missing from assignee xuchenhan-tri(platform), needs at least two assigned reviewers

---

-- commits line 6 at r8:
BTW, I assume RobotLocomotion/drake-external-examples#348 provides test coverage for this?

Code quote:

  The WORKSPACE.bzlmod is no longer load-bearing, so as of this commit
  it should be possible to consume Drake as a module external of other
  downstream projects (instead of as a repository external).

---

WORKSPACE.bzlmod line 5 at r8 (raw file):

# This file lists Drake's workspace-style external dependencies that are only
# needed by Drake Developers instead of downstream projects. Most dependencies
# are listed in MODULE.bazel, instead.

BTW, IIUC, the intent is that this file should only shrink, but never grow, in size. Consider making that point clearer.

Suggestion:

# needed by Drake Developers instead of downstream projects. All other dependencies
# are listed in MODULE.bazel, instead.

[jwnimmer-tri]

Reviewable status: 2 unresolved discussions, LGTM missing from assignee xuchenhan-tri(platform), needs at least two assigned reviewers

---

-- commits line 6 at r8:

Previously, xuchenhan-tri wrote…

BTW, I assume RobotLocomotion/drake-external-examples#348 provides test coverage for this?

Yes, exactly right. That's the most relevant / important test of this feature.

To a certain extent, our pre-merge CMake builds also provide coverage because they create a new dummy module ("drake_cmake") which consumes the "drake" module as a bzlmod external.

---

WORKSPACE.bzlmod line 5 at r8 (raw file):

Previously, xuchenhan-tri wrote…

BTW, IIUC, the intent is that this file should only shrink, but never grow, in size. Consider making that point clearer.

Working

In fact, I should probably take it a step further and make this file only for CLion Drake Developers and disable off the workspace entirely in our default rcfile.

I will push on that tomorrow.

[jwnimmer-tri]

Reviewed 4 of 4 files at r10, all commit messages.
Reviewable status: 1 unresolved discussion, LGTM missing from assignee xuchenhan-tri(platform), needs at least two assigned reviewers

---

WORKSPACE.bzlmod line 5 at r8 (raw file):

Previously, jwnimmer-tri (Jeremy Nimmer) wrote…

Working

In fact, I should probably take it a step further and make this file only for CLion Drake Developers and disable off the workspace entirely in our default rcfile.

I will push on that tomorrow.

Done

[xuchenhan-tri]

Feature

Reviewed 3 of 17 files at r1, 1 of 10 files at r2, 1 of 7 files at r3, 2 of 18 files at r7, 6 of 18 files at r8, 2 of 4 files at r9, 4 of 4 files at r10, all commit messages.
Reviewable status: 6 unresolved discussions, needs at least two assigned reviewers

---

MODULE.bazel line 32 at r10 (raw file):

    "cc_configure_extension",
)
use_repo(cc_configure, "local_config_cc")

BTW, where and how is the local_config_cc repo used now that the cc external is deprecated?

---

MODULE.bazel line 88 at r10 (raw file):

    internal_repositories,
    "abseil_cpp_internal",
    "bazelisk",

BTW, I realize that some of these repos don't have the suffix _internal. Do we have plans to make the names consistent?

---

tools/install/libdrake/header_lint.bzl line 15 at r10 (raw file):

    "+drake_dep_repositories+eigen",
    "+drake_dep_repositories+fmt",
    "+drake_dep_repositories+lcm",

BTW, I don't see lcm in the header_dependency_test. Should it be added?

Also, whoever came up with this spelling with +'s definitely managed to trip me over it everytime I look at it.

---

tools/workspace/default.bzl line 121 at r10 (raw file):

    """Declares workspace repositories for all externals needed by drake (other
    than those built into Bazel, of course).  This is intended to be loaded and
    called from a WORKSPACE file.

BTW, this is now also called in the implementation of the module extension.

Code quote:

    than those built into Bazel, of course).  This is intended to be loaded and
    called from a WORKSPACE file.

---

tools/workspace/workspace_bzlmod_sync_test.py line 144 at r10 (raw file):

    def test_default_exported_sync(self):
        """Our default.bzl has a list of REPOS_EXPORTED that must match the

BTW, what about internal repos? Do we want to check that the repos used through MODULE and WORKSPACE match? Or is the fact that things compile is good enough?

---

tools/workspace/pybind11/BUILD.bazel line 20 at r10 (raw file):

# This alias provides a single point of control for defining which pybind11
# library our tools/skylark/pybind.bzl macro should use.
alias(

BTW, how do we be sure that this is the only symbol that causes issue when switching to bzlmod? Through tests in drake_external_examples?

[jwnimmer-tri]

Reviewed 3 of 3 files at r11, all commit messages.
Reviewable status: 3 unresolved discussions, needs at least two assigned reviewers

---

MODULE.bazel line 32 at r10 (raw file):

Previously, xuchenhan-tri wrote…

BTW, where and how is the local_config_cc repo used now that the cc external is deprecated?

The effect of this call is to register a C++ toolchain with Bazel, which ends up being the highest priority toolchain and thus implicitly used by rules like cc_library.

---

MODULE.bazel line 88 at r10 (raw file):

Previously, xuchenhan-tri wrote…

BTW, I realize that some of these repos don't have the suffix _internal. Do we have plans to make the names consistent?

Added a TODO. My thinking it to not really worry about it until after we've deprecate WORKSPACE mode entirely.

---

tools/install/libdrake/header_lint.bzl line 15 at r10 (raw file):

Previously, xuchenhan-tri wrote…

BTW, I don't see lcm in the header_dependency_test. Should it be added?

Also, whoever came up with this spelling with +'s definitely managed to trip me over it everytime I look at it.

Explained with new comment.

---

tools/workspace/workspace_bzlmod_sync_test.py line 144 at r10 (raw file):

Previously, xuchenhan-tri wrote…

BTW, what about internal repos? Do we want to check that the repos used through MODULE and WORKSPACE match? Or is the fact that things compile is good enough?

My belief is that for internal repos, any kind of discrepancy will immediately result in a build error, so we don't need a linter.

---

tools/workspace/pybind11/BUILD.bazel line 20 at r10 (raw file):

Previously, xuchenhan-tri wrote…

BTW, how do we be sure that this is the only symbol that causes issue when switching to bzlmod? Through tests in drake_external_examples?

Yes. Testing with DEE and then also with Anzu.

Our https://drake.mit.edu/stable.html API forbids people from using nearly all of our skylark macros in the first place (other than repository rules), so really nobody was supposed to be subject to being broken here anyway. However we had a few place in DEE and Anzu that were using pybind.bzl outside of the Stable API, so it seemed easy enough to keep that working.

Sometime later this year I hope to pull pybind11 from BCR anyway (if we're still using it) and there it comes along with a ruleset already, so hopefully this all becomes moot.

[rpoyner-tri]

+@rpoyner-tri for platform review.

Reviewable status: 3 unresolved discussions, LGTM missing from assignee rpoyner-tri(platform)

[xuchenhan-tri]

Reviewed 3 of 3 files at r11, all commit messages.
Reviewable status: LGTM missing from assignee rpoyner-tri(platform)

[rpoyner-tri]

checkpoint -- more later.

Reviewed 2 of 2 files at r5, 1 of 18 files at r7, 7 of 18 files at r8, 1 of 4 files at r9, 1 of 3 files at r11, all commit messages.
Reviewable status: 3 unresolved discussions, LGTM missing from assignee rpoyner-tri(platform)

---

MODULE.bazel line 217 at r11 (raw file):

# TODO(#20731) More improvements are still needed to our MODULE organization:
# - Switch public API dependencies (e.g., eigen) to use modules.
# - Provide better configuation options for choosing dependencies.

typo

Suggestion:

configuration

---

tools/workspace/default.bzl line 115 at r11 (raw file):

# =============================================================================
# For Bazel projects using Drake as a depedency via the WORKSPACE mechanism.

typo

Suggestion:

dependency

---

tools/workspace/default.bzl line 397 at r11 (raw file):

# =============================================================================
# For Bazel projects using Drake as a depedency via the MODULE mechanism.

typo

Suggestion:

dependency

[rpoyner-tri]

Reviewed 1 of 18 files at r8, 1 of 4 files at r10.
Reviewable status: 3 unresolved discussions, LGTM missing from assignee rpoyner-tri(platform)

[rpoyner-tri]

Reviewed 2 of 17 files at r1, 3 of 10 files at r2, 1 of 7 files at r3, 2 of 18 files at r7, 5 of 18 files at r8, 2 of 4 files at r9, 3 of 4 files at r10, 2 of 3 files at r11.
Reviewable status: 3 unresolved discussions

[jwnimmer-tri]

Reviewed 2 of 2 files at r12, all commit messages.
Reviewable status: complete! all discussions resolved, LGTM from assignees rpoyner-tri(platform),xuchenhan-tri(platform)

[rpoyner-tri]

Reviewed 2 of 2 files at r12, all commit messages.
Reviewable status: complete! all discussions resolved, LGTM from assignees rpoyner-tri(platform),xuchenhan-tri(platform)