You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
I want input on this one. There is currently a mix of .h and .hpp (most things have .h). I want to resolve this once and have a consistent naming moving forward. This needs to be resolved reletively soon, as we really need to get the code into shape for external usage.
Option 1: all c++ headers have .hpp extension and all c headers have .h extension
Option 2: all headers (c/c++) have .h extension
I think option 1 is long-term a bit cleaner, but may break more downstream code.
I was reading some discussions on it online (especially this CppCoreGuidelines issue). In my opinion, since PCMS is intended to provide a specific C and Fortran API, I would go for Option 1 for users' convenience. Even if it breaks some code now, it will be cleaner once and for all.
I also think that no other extension headers should be allowed, especially the interface headers. For example, .tpp is sometimes used for templated code.
@Sichao25@Fuad-HH@abhiyan123@hgangwar I'd like to take care of this in one fell swoop. Potentially a rename commit immediately after the mesh fields breaking change? I think the current approach where I've been doing it little by little is probably worse than just doing it all at once.
Activity
Fuad-HH commentedon Mar 8, 2026
I was reading some discussions on it online (especially this CppCoreGuidelines issue). In my opinion, since PCMS is intended to provide a specific C and Fortran API, I would go for Option 1 for users' convenience. Even if it breaks some code now, it will be cleaner once and for all.
I also think that no other extension headers should be allowed, especially the interface headers. For example,
.tppis sometimes used for templated code.jacobmerson commentedon Mar 8, 2026
I agree about the
.tppfiles. If nothing else, it breaks code highlighting / confuses the IDEs.jacobmerson commentedon Aug 14, 2026
@Sichao25 @Fuad-HH @abhiyan123 @hgangwar I'd like to take care of this in one fell swoop. Potentially a rename commit immediately after the mesh fields breaking change? I think the current approach where I've been doing it little by little is probably worse than just doing it all at once.