Is it okay to assume that the traits.pdf in the Github page is the latest document?
(2025-06) Also, I wonder whether the proposal PDF has been analyzed extensively by the latest LLMs for potential/predictable issues at this stage (before full compiler implementation)…?
No, we haven’t updated it yet on Github, and I’d think it will still take some time for us to do so, because we always perform a final internal review before we upload, and @certik is presently too occupied with LFortran bug fixing.
We could also have it analyzed more extensively by LLMs in the process. We’ve already explored aspects of its implementation in this manner, along with aspects of future KART genericity support.
1 Like
I think @kkifonidis’s proposal might be the last proposal written without any AI. So the quality should be high. 
2 Likes
Are they going to be in the next standard revision, or we may have to wait longer? Or will LFortran be the spearhead?
LFortran will definitely prototype these features to give everyone the chance to try them out. With further demand, I am sure they would also find their way (mainly via community support) into other open source compilers, especially Flang.
Help by hardware vendors would also be appreciated. But it’s not a prerequisite, since LFortran and Flang are both open source, and are in the process of becoming the de facto standard Fortran compilers for the future (with Flang already making up the Fortran frontend of most hardware vendors’ compilers).
Both of these compilers are based on LLVM and its back ends, and LLVM is again supported by all hardware vendors. Hence, Fortran codes that will use these features will be portable.
EDIT: I’ll leave it to @certik to comment on LFortran’s implementation time-line of these features.
We are ready to implement it, but @kkifonidis and I thought maybe let’s get to beta first, fixing most common bugs and get most codes working, to provide a solid foundation for further work, so I’ve been focusing on that.
2 Likes