Automate updates

[dcooley]

Some sort of Bash script to:

• clone the latest version of uber/h3 repo
• remove all the src/ files
• rebuild src/init.c
• Report on the differences between the new clone and the outgoing version
• rebuild inst/include/h3libapi.h

[RayShao01010010]

I'm really sorry but I forgot the switch to branch issue4 before committing and pushing the changes.

Now working on branch issue4

[RayShao01010010]

This is now finished and ready for review

[RayShao01010010]

Did some testing on the automation script and found tons of issues. Fixed them all and now the script is working.

There is still one issue though. After the update, the result of getHexagonEdgeLengthAvgKm and getHexagonEdgeLengthAvgM have been changed.

I noticed that they have changed the data matrix in those functions, meaning that maybe we should update our unit tests.

> test_all(pkgdir = "./", testdir = "./tests/tinytest")
test_directed_edges.R.........    8 tests OK 6ms
test_hierarchy.R..............    7 tests OK 6ms
test_indexing.R...............    3 tests OK 9ms
test_inspection.R.............    7 tests OK 7ms
test_miscellaneous.R..........   20 tests 2 fails 19ms
3ms
test_traversal.R..............    7 tests OK 5ms
test_vertexes.R...............    4 tests OK 4ms
----- FAILED[data]: test_miscellaneous.R<9--9>
 call| expect_equal(getHexagonEdgeLengthAvgKm(res = c(5, 6)), c(8.544408276, 
 call| -->    3.229482772))
 diff| Mean relative difference: 0.1532826
 ----- FAILED[data]: test_miscellaneous.R<10--10>
 call| expect_equal(getHexagonEdgeLengthAvgM(res = c(5, 6)), c(8544.408276, 
 call| -->    3229.482772))
 diff| Mean relative difference: 0.1532826
 
Showing 2 out of 56 results: 2 fails, 54 passes (72ms)
> getHexagonEdgeLengthAvgKm(res = c(5L, 6L))
[1] 9.854091 3.724533
> getHexagonEdgeLengthAvgM(res = c(5L, 6L))
[1] 9854.091 3724.533

Anyway, I'm happy this is working on my end. Please test it out

[dcooley]

Those failing tests for getHexagonEdgeLengthAvgKm/m() show the distances has changed by about 1km at res 5, which may or may not be an issue. But there's nothing in the h3 changelog about this.

So I think we have to adjust the test to use the new results

[dcooley]

I've adjusted the tests in {h3r}

[dcooley]

merged

[dcooley]

Had to make some changes to get on CRAN

• Various functions with empty arguments need void a the parameter (e.g. DECLSPEC int H3_EXPORT(res0CellCount)(void);)
• h3Assert.c needed a definition of something (typedef int unused;)
• init.c has 2 non-api functions

[dcooley]

closing because I think #7 superseeds this now