On Thu, Dec 19, 2024 at 08:18:38AM +0700, Robert Elz wrote: > Date: Wed, 18 Dec 2024 19:45:36 +0100 > From: Roland Illig <roland.il...@gmx.de> > Message-ID: <c9cd653a-3ece-4cdb-a5dc-7e380db1f...@gmx.de> > > | > You need to first test for EOF but then cast any non-EOF characters to > | > unsigned char to be portable: > | > | That's wrong. The CAVEATS section explicit allows character I/O > operations: > > > Yes, passing EOF (as an int) to these macros/functions is permitted, there > is no need to test for it as Martin's sample code did. > > The problem case is when the return from [f]getc() is stored into a > char type variable, and then that variable is passed to the <ctype.h> > macros/functions - that doesn't work. As long as only int variables > are used, it is all OK.
The problem was that it was stored in an unsigned int. uint32_t c = EOF; isspace(c); -> segmentation fault on nbsd Cheers, Patrick
