[workspace] Add vtk_internal build from source

[jwnimmer-tri]

Towards #19945 and #16502.

---

This change is 

[jwnimmer-tri]

@drake-jenkins-bot mac-x86-monterey-clang-bazel-experimental-release please

[jwnimmer-tri]

+@rpoyner-tri for feature review, please.

[rpoyner-tri]

I haven't thought about TGA files in literally decades.

Reviewed 12 of 12 files at r1, all commit messages.
Reviewable status: 7 unresolved discussions, needs at least two assigned reviewers

---

tools/workspace/vtk_internal/repository.bzl line 47 at r1 (raw file):

        key, values = tokens[0], tokens[1:]
        if key.endswith("S"):
            # Plural key.

This test happens to work with the keys documented, but the overall effect is somewhat misleading.

The distinction in number is nothing to do with the incidental grammar status of the keys; its rather whether the key takes one or multiple values.

Perhaps instead:

multivalued_keys = ["DEPENDS", "PRIVATE_DEPENDS"]  # maybe others later?
if key in multivalued_keys:
    # Multi-valued key.
   # ... blah
else:
    # Single-valued key.

Or, if you are confident that no = characters will ever be embedded in values, just branch on the length of the token list?

---

tools/workspace/vtk_internal/repository.bzl line 65 at r1 (raw file):

    an loadable `modules.bzl` file in the root of the repository.

    This is necessary because BUILD files can't parse external metdata as part

typo

Suggestion:

metadata

---

tools/workspace/vtk_internal/rules.bzl line 59 at r1 (raw file):

            module_deps_private.remove(dep_name)

    # Generate the module header, ala CMake GenerateExportHeader.

typo

Suggestion:

a la

---

tools/workspace/vtk_internal/rules.bzl line 75 at r1 (raw file):

            ]).format(scream = scream),
        )
        hdrs_extra = hdrs_extra + [module_h]

nit does not hdrs_extra.append(module_h) work in Starlark?

At least one other instance below.

---

tools/workspace/vtk_internal/rules.bzl line 92 at r1 (raw file):

    ] + hdrs_glob_exclude)
    gen_hdrs = [
        "gen/" + x[:-3]

This slice is a masterpiece of obfuscation. What exactly is the goal here?

---

tools/workspace/vtk_internal/rules.bzl line 186 at r1 (raw file):

            build this module, typically loaded from a `settings.bzl` file.

    Keys of module_metaata:

typo (or https://www.youtube.com/watch?v=nbY_aP-alkw ?)

Suggestion:

metadata

---

tools/workspace/vtk_internal/rules.bzl line 309 at r1 (raw file):

def generate_abi_namespace():
    """Generates vtkABINamespace.h, mimicing the Common/Core/CMakeLists.txt.

typo

Suggestion:

mimicking

[jwnimmer-tri]

Reviewable status: 3 unresolved discussions, needs at least two assigned reviewers

---

tools/workspace/vtk_internal/repository.bzl line 47 at r1 (raw file):

Previously, rpoyner-tri (Rick Poyner (rico)) wrote…

This test happens to work with the keys documented, but the overall effect is somewhat misleading.

The distinction in number is nothing to do with the incidental grammar status of the keys; its rather whether the key takes one or multiple values.

Perhaps instead:

multivalued_keys = ["DEPENDS", "PRIVATE_DEPENDS"]  # maybe others later?
if key in multivalued_keys:
    # Multi-valued key.
   # ... blah
else:
    # Single-valued key.

Or, if you are confident that no = characters will ever be embedded in values, just branch on the length of the token list?

Possibly clearer now?

---

tools/workspace/vtk_internal/rules.bzl line 75 at r1 (raw file):

Previously, rpoyner-tri (Rick Poyner (rico)) wrote…

nit does not hdrs_extra.append(module_h) work in Starlark?

At least one other instance below.

Nope. Function arguments are immutable (readonly). If I wanted to append, I'd need to copy it first.

---

tools/workspace/vtk_internal/rules.bzl line 92 at r1 (raw file):

Previously, rpoyner-tri (Rick Poyner (rico)) wrote…

This slice is a masterpiece of obfuscation. What exactly is the goal here?

Possibly clearer now?

[jwnimmer-tri]

+@SeanCurtis-TRI for platform review, please. Rico will be on-call tomorrow, and you're a good candidate for "a developer who might need to understand this stuff enough to make a small adjustment" so my goal is for you to be able to find your way around here.

[rpoyner-tri]

Reviewed 2 of 2 files at r2, all commit messages.
Reviewable status: LGTM missing from assignee SeanCurtis-TRI(platform)

[SeanCurtis-TRI]

With a couple of typos.

Reviewed 10 of 12 files at r1, 2 of 2 files at r2, all commit messages.
Reviewable status: 4 unresolved discussions

---

tools/workspace/vtk_internal/repository.bzl line 66 at r2 (raw file):

            result[key] = 1
        else:
            fail("{}: Too may values for {}".format(subdir, key))

nit: typo

Suggestion:

many

---

tools/workspace/vtk_internal/repository.bzl line 66 at r2 (raw file):

            result[key] = 1
        else:
            fail("{}: Too may values for {}".format(subdir, key))

BTW It seems that this error message is a regression against the trailing "S" heuristic. The defect may not be that there are too many values, but that there are multiple values, but we weren't told to interpret it as a list. Is it worth rephrasing the error message accordingly?

---

tools/workspace/vtk_internal/repository.bzl line 71 at r2 (raw file):

def create_modules_bzl(repo_ctx):
    """Finds all vtk.module files, parses them, and writes their content into
    an loadable `modules.bzl` file in the root of the repository.

nit: typo

Suggestion:

a 

---

tools/workspace/vtk_internal/patches/io_image_formats.patch line 28 at r2 (raw file):

   reader->Delete();
-  vtkImageReader2Factory::AvailableReaders->AddItem((reader = vtkHDRReader::New()));
+  // vtkImageReader2Factory::AvailableReaders->AddItem((reader = vtkHDRReader::New()));

BTW This will be coming back soon, because we'll want to set the envirornment map in RenderEngineVtk via .hdr files. You don't need to take any action now, I'm just flagging it to help log this into my own memory.

[jwnimmer-tri]

Reviewable status: complete! all discussions resolved, LGTM from assignees rpoyner-tri(platform),SeanCurtis-TRI(platform)

---

tools/workspace/vtk_internal/repository.bzl line 66 at r2 (raw file):

Previously, SeanCurtis-TRI (Sean Curtis) wrote…

BTW It seems that this error message is a regression against the trailing "S" heuristic. The defect may not be that there are too many values, but that there are multiple values, but we weren't told to interpret it as a list. Is it worth rephrasing the error message accordingly?

Done

[SeanCurtis-TRI]

Reviewed 2 of 2 files at r3, all commit messages.
Reviewable status: complete! all discussions resolved, LGTM from assignees rpoyner-tri(platform),SeanCurtis-TRI(platform)