Re: lib/53640: getnstr(3) shall read at most n characters, not n-1

[Valery Ushakov <uwe%stderr.spb.ru@localhost>]

The following reply was made to PR lib/53640; it has been noted by GNATS.

From: Valery Ushakov <uwe%stderr.spb.ru@localhost>
To: gnats-bugs%NetBSD.org@localhost
Cc: 
Subject: Re: lib/53640: getnstr(3) shall read at most n characters, not n-1
Date: Sat, 29 Sep 2018 18:29:35 +0300

 On Sat, Sep 29, 2018 at 14:05:00 +0000, Kamil Rytarowski wrote:
 
 >  >  On Sat, Sep 29, 2018 at 10:30:01 +0000, n54%gmx.com@localhost wrote:
 >  >
 >  >  > It's documented in POSIX to reat at most N bytes; but the NetBSD
 >  >  > curses(3) version reads at most N-1 bytes.
 >  >  >
 >  >  > http://pubs.opengroup.org/onlinepubs/7908799/xcurses/getnstr.html
 >  >
 >  >  I don't think that definition is precise enough.  It doesn't
 >  >  explicitly say if the buffer is NUL terminated on overflow.
 >  
 >  It's specified that the input string is terminated by a \0 byte.
 >  
 >  "The resulting value is placed in the area pointed to by str. The string
 >  is then terminated with a null byte. "
 >  
 >  In combination of "functions read at most n bytes" I assume that I can
 >  enter n characters (bytes) and a string is always terminated with a NUL
 >  character.
 
 Given the overall confusion that usually surrounds these kinds of "n"
 functions, it would be better that this kind of details is specified
 as explicitly as possible.
 
 E.g. another wording issue to pick on - does the newline or carriage
 return 1) count towards "N" and 2) are they stored?
 
 Solaris manual page explicitly says newline IS placed into the
 destination string:
 
      These  functions  call  wgetch(3XCURSES)  and   place   each
      received character in str until a newline is received, which
      is also placed in str.
 
 but it fact tests show that it's NOT stored.
 
 ncurses explicitly says the newline is NOT placed into the output
 buffer.
 
 And so, between all possible permutations of what is said and what is
 not and what of what is said is actually wrong there are so many
 possible ways to interpret this.
 
 As I said in my previous reply, all manual pages for this suck in one
 way or another and I still think NetBSD behavior seems globally
 correct (POSIX wording notwithstanding).
 
 
 >  >  Is it really that hard to explicitly quote the results you observe
 >  >  with these tests?  I find it rather disingenuous.  Can't you
 >  >  explicitly state your point?
 >  
 >  This is an interactive program. You cannot enter N characters into it,
 >  only up to N-1.
 
 *sigh*  And again you are speaking in the abstract.  Not about the
 specific test code you quoted.
 
 
 >  -----BEGIN PGP SIGNATURE-----
 
 Please, can you turn this off for gnats replies?  Combined with
 quoted-printable of the actual reply it clutters the bug report,
 making it harder to read.
 
 
 -uwe