• POSIX: printf ---?

    From Christian Weisgerber@3:633/10 to All on Tuesday, November 17, 2026 17:43:26
    Any POSIX lawyers around?

    Various printf(1) implementations attempt option parsing on the first
    argument and choke on something like "printf ---":

    FreeBSD sh(1) builtin: printf: Illegal option --
    FreeBSD standalone: printf: illegal option -- -
    bash(1) builtin: -bash: printf: --: invalid option

    POSIX does not specify any options for printf(1).

    The workaround is to explicitly terminate option parsing with "--",
    i.e., "printf -- ---". That works at least across OpenBSD, FreeBSD,
    GNU coreutils and bash. But does it conform to POSIX?


    The problem isn't hypothetical. In release 0.131, Game of Tree's
    regression test suite added things like

    + printf '--- gamma/delta\t\n' >> $testroot/stdout.expected
    + printf '+++ gamma/delta\t\n' >> $testroot/stdout.expected

    which promptly blew up on FreeBSD and presumably some flavors of
    Linux.

    --
    Christian "naddy" Weisgerber naddy@mips.inka.de

    --- PyGate Linux v1.5.19
    * Origin: Dragon's Lair, PyGate NNTP<>Fido Gate (3:633/10)
  • From Kaz Kylheku@3:633/10 to All on Tuesday, November 17, 2026 20:46:28
    On 2026-11-17, Christian Weisgerber <naddy@mips.inka.de> wrote:
    Any POSIX lawyers around?

    I'm supposed to be in court, but it was unexpectedly adjourned.

    Various printf(1) implementations attempt option parsing on the first argument and choke on something like "printf ---":

    There is a "--" argument which means "this is the last option; only
    ordinary non-option arguments follow". Obviously, "---" is not that
    argument.

    FreeBSD sh(1) builtin: printf: Illegal option --
    FreeBSD standalone: printf: illegal option -- -
    bash(1) builtin: -bash: printf: --: invalid option

    POSIX does not specify any options for printf(1).

    However, POSIX does still specify the -- option for printf; it's just
    not spelled out in the description of printf.

    Contrast this situation with that of the echo utility. For echo,
    much more is written than just "OPTIONS ... None":

    The echo utility shall not recognize the "--" argument in the manner
    specified by Guideline 10 of XBD 12.2 Utility Syntax Guidelines; "--"
    shall be recognized as a string operand.

    Implementations shall not support any options.

    The printf utility is in no way similarly excused from supporting --,
    nor forbidden from having options courtesy of the implementor.

    (Implementations of echo, such as the Bash builtin and the coreutils
    echo, ignore the proscription against supporting options.)

    The workaround is to explicitly terminate option parsing with "--",
    i.e., "printf -- ---". That works at least across OpenBSD, FreeBSD,
    GNU coreutils and bash. But does it conform to POSIX?

    I believe yes; though POSIX does not specify any required options
    for printf, the utility conforms to the common guidelines.

    The description of printf itself has a FUTURE DIRECTIONS section where
    it is spelled out that implementations can add options:

    A future version of this standard is expected to add a %b conversion to
    the printf() function for binary conversion of integers, in alignment
    with the next version of the ISO C standard. This will result in an
    inconsistency between the printf utility and printf() function for
    format strings containing %b. Implementors are encouraged to collaborate
    on a way to address this which could then be adopted in a future version
    of this standard. For example, the printf utility could add a -C option
    to make the format string behave in the same way, as far as possible, as
    the printf() function.

    printf is allowed to treat arguments that start with - as options,
    and diagnose nonexistent options.

    Highly portable scripts must take care not to use format strings
    that start with -.

    Also, do not pass the expansion of a variable to printf:

    printf "$TEXT\n" # no

    printf "%s\n" "$TEXT" # yes

    --
    TXR Programming Language: http://nongnu.org/txr
    Cygnal: Cygwin Native Application Library: http://kylheku.com/cygnal
    Mastodon: @Kazinator@mstdn.ca

    --- PyGate Linux v1.5.19
    * Origin: Dragon's Lair, PyGate NNTP<>Fido Gate (3:633/10)
  • From Lawrence D?Oliveiro@3:633/10 to All on Tuesday, November 17, 2026 21:37:41
    On Tue, 17 Nov 2026 17:43:26 -0000 (UTC), Christian Weisgerber wrote:

    Various printf(1) implementations attempt option parsing on the first argument and choke on something like "printf ---":

    FreeBSD sh(1) builtin: printf: Illegal option --
    FreeBSD standalone: printf: illegal option -- -
    bash(1) builtin: -bash: printf: --: invalid option

    The GNU coreutils version prints the ?---? without error.

    The workaround is to explicitly terminate option parsing with "--",
    i.e., "printf -- ---". That works at least across OpenBSD, FreeBSD,
    GNU coreutils and bash. But does it conform to POSIX?

    Or how about

    printf %s ---

    Or just use echo and be done with it.

    --- PyGate Linux v1.5.19
    * Origin: Dragon's Lair, PyGate NNTP<>Fido Gate (3:633/10)
  • From Geoff Clare@3:633/10 to All on Wednesday, November 18, 2026 13:21:43
    Kaz Kylheku wrote:

    On 2026-11-17, Christian Weisgerber <naddy@mips.inka.de> wrote:

    Various printf(1) implementations attempt option parsing on the first
    argument and choke on something like "printf ---":

    FreeBSD sh(1) builtin: printf: Illegal option --
    FreeBSD standalone: printf: illegal option -- -
    bash(1) builtin: -bash: printf: --: invalid option

    POSIX does not specify any options for printf(1).

    However, POSIX does still specify the -- option for printf; it's just
    not spelled out in the description of printf.

    Contrast this situation with that of the echo utility. For echo,
    much more is written than just "OPTIONS ... None":

    The echo utility shall not recognize the "--" argument in the manner
    specified by Guideline 10 of XBD 12.2 Utility Syntax Guidelines; "--"
    shall be recognized as a string operand.

    Implementations shall not support any options.

    The printf utility is in no way similarly excused from supporting --,
    nor forbidden from having options courtesy of the implementor.

    Just to back this up with evidence, the requirement for printf derives
    from XCU 1.4 Utility Description Defaults under OPTIONS:

    Default Behavior: When this section is listed as "None.", it means
    that the implementation need not support any options. Standard
    utilities that do not accept options, but that do accept operands,
    shall recognize "--" as a first argument to be discarded.

    The requirement for recognizing "--" is because conforming
    applications need a way to shield their operands from any
    arbitrary options that the implementation may provide as an
    extension. For example, ...

    https://pubs.opengroup.org/onlinepubs/9799919799/utilities/V3_chap01.html#tag_18_04

    (Implementations of echo, such as the Bash builtin and the coreutils
    echo, ignore the proscription against supporting options.)

    I believe they conform to POSIX in this respect. Their documentation
    may well call them options, but what they are allowed to do by POSIX
    is dependent on the contents of the first argument (which according to
    POSIX is an operand):

    The following operands shall be supported:

    string
    A string to be written to standard output. If the first
    operand consists of a '-' followed by one or more characters
    from the set {'e', 'E', 'n'}, or if any of the operands
    contain a <backslash> character, the results are
    implementation-defined.

    (This is followed by some XSI-shaded text requiring that implementations supporting the XSI option always treat string operands as a string to
    be written, and specifying how backslash-escapes are processed.)

    Since only e, E and n produce implementation-defined behaviour,
    attempting to use other "options" should cause the string to be
    written, and this seems to be what they do:

    bash$ echo -x foo
    -x foo
    bash$ /bin/echo -x foo
    -x foo

    Although coreutils printf accepts long options when you don't have POSIXLY_CORRECT set:

    bash$ echo --version
    --version
    bash$ POSIXLY_CORRECT=1 /bin/echo --version
    --version
    bash$ /bin/echo --version
    echo (GNU coreutils) 9.1
    Copyright (C) 2022 Free Software Foundation, Inc.
    License GPLv3+: GNU GPL version 3 or later <https://gnu.org/licenses/gpl.html>. This is free software: you are free to change and redistribute it.
    There is NO WARRANTY, to the extent permitted by law.

    Written by Brian Fox and Chet Ramey.

    --
    Geoff Clare <netnews@gclare.org.uk>

    --- PyGate Linux v1.5.19
    * Origin: Dragon's Lair, PyGate NNTP<>Fido Gate (3:633/10)
  • From Geoff Clare@3:633/10 to All on Wednesday, November 18, 2026 15:57:51
    I wrote:

    Although coreutils printf accepts long options when you don't have

    I've no idea how I ended up writing "coreutils printf" here instead
    of "coreutils echo". Brains are weird.

    POSIXLY_CORRECT set:

    bash$ echo --version
    --version
    bash$ POSIXLY_CORRECT=1 /bin/echo --version
    --version
    bash$ /bin/echo --version
    echo (GNU coreutils) 9.1

    --
    Geoff Clare <netnews@gclare.org.uk>

    --- PyGate Linux v1.5.19
    * Origin: Dragon's Lair, PyGate NNTP<>Fido Gate (3:633/10)