Removing AlignConsecutiveAssignments: true.

[rwgk]

Based on discussion in 2021-06-29 meeting between @EricCousineau-TRI, @henryiii, @rwgk, @Skylion007.

Proposed by rwgk, based on experience clang-formatting the smart_holder code:

• Removal of AlignConsecutiveAssignments.

Motivation: Avoid git diff whitespace clutter. See example below.

This also brings us closer to the pure LLVM style.

While we're at it: also cleaning out a commented-out override. (If we really want it in the future it's easy enough to add back. In the meantime the config file is cleaner.)

Note: rwgk will reformat the smart_holder code after this PR is merged. Currently the master branch is not directly affected, although this PR is meant to be a preparation for globally clang-formatting all pybind11 code (after the next release).

See also: https://github.com/pybind/pybind11/wiki/Roadmap

$ git diff
diff --git a/dummy.cpp b/dummy.cpp
index fd7d1ad8..279eb5ea 100644
--- a/dummy.cpp
+++ b/dummy.cpp
@@ -1,5 +1,6 @@
-int value1 = 1;
-int value2 = 2;
-int value3 = 3;
-int value4 = 4;
-int value5 = 5;
+int value1        = 1;
+int value2        = 2;
+int value3        = 3;
+int value4        = 4;
+int value5        = 5;
+int value_special = 99;

[EricCousineau-TRI]

Just curious - what's the functional change here? Shouldn't this change involve formatting change as in #3065? (sorry for being slow!)

[rwgk]

Just curious - what's the functional change here? Shouldn't this change involve formatting change as in #3065? (sorry for being slow!)

I want to be sure we agree on the format before making changes on the smart_holder branch, by way of having the agreed-on format on master first, even though it's currently not applied here.

What I had in mind for after we agree:

• I will merge this PR into smart_holder.
• Reformat all smart_holder code (essentially PR Changing .clang-format AlignConsecutiveAssignments to false and applying to smart_holder code. #3065).
• Wait for the 2.7 release + a few days.
• Then reformat all pybind11 code on master + add clang-format to the GHA Format workflow.
• Update the smart_holder branch again to also enforce clang-format in the GHA Format workflow.

From then on all pybind11 code will be clang-format clean, enforced automatically.

[rwgk]

Interesting CI flake: one of the two AppVeyor jobs stalled in the middle of running the Python unit tests and then timed out after one hour. I've not seen this before. In any case, it's certainly unrelated to the .clang-format change!

[rwgk]

Hi @EricCousineau-TRI based on your comment under #3065 (#3065 (comment)) I'll go ahead merging this PR.

Next steps:

• I'll apply this to the smart_holder branch (probably via Changing .clang-format AlignConsecutiveAssignments to false and applying to smart_holder code. #3065).
• I'll open an experimental PR to apply this to master, for us to see what the clang-format result looks like.