Re: bash fails to compile on systems with select() but not pselect()

On 2021-05-22 17:56, Koichi Murase wrote: 2021年5月23日(日) 5:08 Larkin Nickle : [...] Notice that `set` is only defined if HAVE_PSELECT is defined. Does it compile in Tru64 by declaring "set/oset" also for HAVE_SELECT? diff --git a/lib/sh/input_avail.c b/lib/sh/input_avail.c index 695165fd..2ac

Re: bash fails to compile on systems with select() but not pselect()

2021年5月23日(日) 5:08 Larkin Nickle : > [...] > > Notice that `set` is only defined if HAVE_PSELECT is defined. Does it compile in Tru64 by declaring "set/oset" also for HAVE_SELECT? diff --git a/lib/sh/input_avail.c b/lib/sh/input_avail.c index 695165fd..2ac44616 100644 --- a/lib/sh/input_avail.c +

bash fails to compile on systems with select() but not pselect()

On systems with select() but without pselect(), bash fails to compile due to what is possibly a bug in `lib/sh/input_avail.c`. This was tested on Tru64 with the bash 5.1 source being built with GCC 4.7.4 on an AlphaServer 1000A. gcc -c -I. -I../.. -I../.. -I../../lib -I../../include -I. -DH

Re: bash fails to compile on systems with select() but not pselect()

Line 145 also contains `sigprocmask (SIG_BLOCK, &set, &oset);` under `#elif defined HAVE_SELECT` even though `set` and `oset` aren't declared unless we have HAVE_PSELECT. #if defined (HAVE_PSELECT) /* XXX - use pselect(2) to block SIGCHLD atomically */ result = pselect (fd + 1, &rea