Test for compacting all res 1 cells

[ajfriend]

• compactCells currently fails when trying to compact all res 1 cells.
• note that, interestingly, it also fails if you try to compact just the first 41 res 1 cells.

[ajfriend]

I'm not sure why it did all this formatting... Any ideas?

[isaacbrodsky]

I'm not sure why it did all this formatting... Any ideas?

Probably different version of clang-format being used?

[ajfriend]

Probably different version of clang-format being used?

Do we have a standard version I should be using?

[isaacbrodsky]

Probably different version of clang-format being used?

Do we have a standard version I should be using?

Yes, it is important to use clang-format-14 (until we update the formatter version again) as they change defaults between versions.

[coveralls]

coverage: 98.825% (+0.001%) from 98.824%
when pulling a3a28e6 on compact_all_res1_test
into 4899d29 on master.

[ajfriend]

Yes, it is important to use clang-format-14 (until we update the formatter version again) as they change defaults between versions.

OK. For my future reference, I had to run

brew install llvm@14
ln -s "$(brew --prefix llvm@14)/bin/clang-format" /opt/homebrew/bin/clang-format

which feels a bit janky.

[ajfriend]

I can also make it fail on a different (disjoint from the first example) set of cells with something like

int64_t numUncompacted = 44;
t_assertSuccess(
    H3_EXPORT(compactCells)(&cells1[80], out, numUncompacted));

And I can get this example to pass if I set numUncompacted = 43.

This is making me wonder if the failure has something to do with how we handle pentagons.

[isaacbrodsky]

The problem is around line

h3/src/h3lib/lib/h3Index.c

Line 490 in b355c9d

return parentError;

, the parentRes is -1 but we try to take cellToParent of currIndex at parentRes anyways.