Add int128 and supported functions

[Ashleyhx]

No description provided.

[acquamarin]

Can you add some test cases to cast_error.test where the casting fails?

[codecov]

Codecov Report

Attention: 67 lines in your changes are missing coverage. Please review.

Comparison is base (5d7ad00) 89.71% compared to head (f2171c0) 89.62%.

Additional details and impacted files

@@            Coverage Diff             @@
##           master    #2096      +/-   ##
==========================================
- Coverage   89.71%   89.62%   -0.09%     
==========================================
  Files        1014     1017       +3     
  Lines       35859    35856       -3     
==========================================
- Hits        32171    32137      -34     
- Misses       3688     3719      +31     

Files	Coverage Δ
src/common/type_utils.cpp	86.41% <100.00%> (+0.34%)	⬆️
src/common/types/types.cpp	0.00% <ø> (-90.83%)	⬇️
src/common/vector/value_vector.cpp	91.96% <100.00%> (+0.12%)	⬆️
src/function/aggregate_function.cpp	88.00% <100.00%> (+0.76%)	⬆️
src/function/comparison_functions.cpp	91.46% <100.00%> (+0.10%)	⬆️
src/function/vector_cast_functions.cpp	81.15% <100.00%> (+0.70%)	⬆️
src/function/vector_hash_functions.cpp	90.47% <100.00%> (+0.47%)	⬆️
src/include/common/exception/overflow.h	100.00% <100.00%> (ø)
src/include/common/string_utils.h	80.00% <ø> (ø)
src/include/common/type_utils.h	100.00% <100.00%> (ø)
... and 33 more

... and 6 files with indirect coverage changes

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

[acquamarin]

Can you also fix the CI and compilation issue on mac?

[Ashleyhx]

@mewim could you take a look at the API part in kuzu.h and in capi/value.cpp?

[acquamarin]

Good job! One general comment is that we should increase the test coverage for edge cases.
I will let @mewim review the API.

[mewim]

I went through the API related changes. The changes for Python and Node.js APIs LGTM.

For C API, the binding is not very user-friendly. The users will not understand val_.low, val_.high until they look into our internal implementation. I suggest to at least bind two more functions: kuzu_int128_t kuzu_int128_t_from_string(const char* str) and char* kuzu_int128_t_to_string(kuzu_int128_t val), so that the user can convert kuzu_int128_t from/to strings without understanding the internal implementation. Also, consider add a TODO item for binding the full set of int128_t functions of the C++ API.

For Java and Rust APIs, I see that only the type enums have been added, but no actual binding implementations. It is fine to implement these bindings as separate PRs, but an issue should be created for them so we don't miss them for the next release.

[mewim]

Also, we should change the title of this PR so it looks better in our release change log.

[acquamarin]

My general comment is to improve test coverage as much as you can.