I can confirm this from testing the release candidates (and earlier) across various packages in our distribution; we’re planning to make flang 19 the default compiler on windows. Congratulations and thanks to the flang developers for this!
All the publicly communicated conditions were met almost a year ago but AFAICT, there remains some undetermined benchmark rate of passed tests from the gfortran test suite to be achieved. Frankly speaking, from the outside this looks & feels like a subversion of the conflict resolution process on this topic, if not in letter then at least in spirit.
This situation forces everyone to keep accumulating technical debt to work around the “temporary” name, and once it is finally changed, it’ll have to be cleaned up again. Build orchestrators like CMake and meson (where I contributed initial support for llvm-flang[1]) need to swallow this complexity, and once fixes finally land, it’ll take years for this to roll out to be widely usable (e.g. available in LTS distributions).
The reason this is grating to me is that I see no benefit that would come close to justifying all that extra work elsewhere. I planned on re-raising this topic in the LLVM 20 timeframe, but the earlier the rename finally happens, the better.
in large part, the reason I didn’t do this already much earlier was precisely that I was waiting for the rename to happen ↩︎