Satish Balay <ba...@mcs.anl.gov> writes: > On Fri, 6 Mar 2020, Jed Brown wrote: >> PetscInt some,several,variables; >> >> // code >> if (PetscDefined(HAVE_MAGIC)) { >> function(several,&variables); >> } >> use(some,variables); > > One minor issue: we haven't yet fixed up clang analyzer build. Likely this > will be listed as 'dead code' block or something like that.
As with #ifdef, accurate code coverage needs to be aggregated across multiple configurations. Are you aware of other problems? >> This approach could also be used to avoid needing separate macros for >> every SETERRQ1-SETERRQ9, etc. I have an example implementation in this >> MR, and it passes the full pipeline (after relaxing the -std=c89 >> -pedantic build). > > These are in public includes. So apps using -std=c89 won't have access to > this part of the API? Oh, I would keep the macros for a while (perhaps with gradual deprecation starting eventually), just define the new ones for supported dialects and use the new ones internally.