Interesting chicken-and-egg prroblem __has_c_attribute is not defined until C23, but then you don't have a mechanism to check for [[deprecated]]. Good thing these are both added in C23, so we can check for both at same time.

Codecov Report

All modified and coverable lines are covered by tests ✅

Project coverage is 83.59%. Comparing base (9ab66b3) to head (974151a).

@@           Coverage Diff            @@
##           develop     #435   +/-   ##
========================================
  Coverage    83.59%   83.59%           
========================================
  Files           24       24           
  Lines         8118     8118           
  Branches      1689     1693    +4     
========================================
  Hits          6786     6786           
  Misses        1332     1332           

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

@LecrisUT, I didn't know about those features in C23. They are simply ignored in older C standard?

@LecrisUT, I didn't know about those features in C23. They are simply ignored in older C standard?

Actually they have been available for quite sometime in the major compilers, check the fallthrough, it covers gnu, clang, and microsoft. Incidentally we can check that they are functioning by looking at the build log of each compiler since we are using those deprecated functions to cover in the test-suite (I should patch it to disable the warnings at some point).

Edit: actually microsoft deprecation is not triggered 🤷

Btw, the macros are quite hard to read, wdyt of imposing an indentation to it:
https://clang.llvm.org/docs/ClangFormatStyleOptions.html#indentppdirectives

I like the BeforeHash format.

Btw, the macros are quite hard to read, wdyt of imposing an indentation to it:

I didn't know that we can indent the macro line. I have no objections and any convention is fine to me.

Btw, the macros are quite hard to read, wdyt of imposing an indentation to it:

I didn't know that we can indent the macro line. I have no objections and any convention is fine to me.

@lan496 Preference?

Let's use PPDIS_BeforeHash!

Note that I only marked random functions. Please go through the others as well to decide if it should be depricated or not

I think I've marked all of the interfaces that can be easily deprecated. In general many of the interfaces should be reworked to reuse the dataset and have more nested interface.

• spg_standardize_cell, spg_find_primitive, etc. should be converted to getters of SpglibDataset
• Reduce the interface where you pass lattice/position/rotation/translation directly
• ...get_symmetry_with_collinear_spin should also be reworked so that it can re-use a pre-calculated dataset "object". Will be easier to do that with a cpp interface, and making the C interface go through a void* pointer
• spg_get_symmetry_from_database feels like it should have its own datatype from which SpglibDataset expands
• I am not sure about the kpoint grid interface