Elbert Pol wrote: > I attach the logs > > This is with os/2 Except for the well-known SIGFPEs (which are most probably caused by computations with NaNs) there are new failures of test-open and test-fopen. config.log has configure:18187: checking whether fopen recognizes a trailing slash configure:18255: result: no configure:23736: checking whether open recognizes a trailing slash configure:23806: result: no #define OPEN_TRAILING_SLASH_BUG 1 #define FOPEN_TRAILING_SLASH_BUG 1 and no FCHDIR_REPLACEMENT make.log compiles open.c and fopen.c.
> I have use now gcc 3.4.6 instead of the 3.3.5 Maybe you also have a working debugger now? Can you single-step through test-open and show the contents of the statbuf after it has been filled by fstat? $ gdb test-open.exe break rpl_open run break 116 continue nexti print statbuf nexti print statbuf nexti print statbuf nexti print statbuf nexti print statbuf nexti print statbuf nexti print statbuf Bruno
