Tom G. Christensen wrote: > > - Does 1.0 / -0.0 < 0 evaluate to true or false? > > - Does 1.0L / -0.0L < 0 evaluate to true or false? > true > true
Thanks. That's different from what I got on some IRIX 6.5 machine, but it's consistent with what you and Nelson see. > > - At test-vasprintf-posix.c:1309, what is the resulting string? > > > On both 5.3 and 6.2: > 0.000000 33 OK, it's really the sign that is missing. > > - When you single-step from test-vasprintf-posix.c:1304 or :1305 on, > > where is the '-' sign dropped? > > > I'm not sure what I should be looking for? > I've attempted to single-step as requested but I might not be printing > the right thing. > This is on Irix 6.2: > dbx version 7.3 MR 55458_Apr30_MR Apr 30 1999 13:44:41 > Executable > /usr/people/tgc/build/athena-buildpkg/buildpkg/m4/src/m4-1.4.11/tests/test-vasprintf-posix > (dbx) stop at 1305 > Process 0: [3] stop at > "/usr/people/tgc/build/athena-buildpkg/buildpkg/m4/src/m4-1.4.11/tests/test-vasprintf-posix.c":1305 > (dbx) run > Process 1047 (test-vasprintf-posix) started > [3] Process 1047 (test-vasprintf-posix) stopped at [test_function:1305 > ,0x10008f3c] > 1305 int retval = > (dbx) print result > (nil) > (dbx) step > Process 1047 (test-vasprintf-posix) stopped at [my_asprintf:3359 > ,0x100146c8] > 3359 va_start (args, format); > (dbx) print result > 0x7ffb66a4 > (dbx) step > Process 1047 (test-vasprintf-posix) stopped at [my_asprintf:3360 > ,0x100146d0] > 3360 ret = vasprintf (result, format, args); > (dbx) print result > 0x7ffb66a4 > (dbx) step > Process 1047 (test-vasprintf-posix) stopped at [vasprintf:37 ,0x100147f4] > 37 char *result = vasnprintf (NULL, &length, format, args); > (dbx) print result > 0x100251f0 = "0.000000 33" > (dbx) step > Process 1047 (test-vasprintf-posix) stopped at [vasnprintf:1416 > ,0x10014c04] > 1416 if (PRINTF_PARSE (format, &d, &a) < 0) > (dbx) Thanks, that's good so far. The interesting stuff is to look - what happens in line vasnprintf.c:2734, - at the result in line vasnprintf.c:4635. (At this point, I don't know whether the wrong result comes from the system's sprintf or snprintf function or is produced by gnulib's vasnprintf.c directly.) Bruno
