US228 20.20.02.2 [format.string.std] p7,9 Specify units of width and precision P1868 LWG 3290

[wg21bot]

Units of width and precision are not specified which causes an ambiguity for strings in variable-length encodings.

Proposed change:
Width and precision for strings should be computed based on fixed operating system dependent encodings. If the operating system is capable of displaying Unicode text in a terminal both ordinary and wide encodings are Unicode encodings such as UTF-8 and UTF-16, respectively. [ Note: this is the case for Windows-based and many POSIX-based operating systems. -- end note ] Otherwise encodings are implementation-defined. For the given encoding, display width of a string is the number of column positions needed to display the string in a terminal [ Note: This is similar to the semantics of the POSIX wcswidth function with a fixed encoding. —]

[jensmaurer]

See also P1868 width: clarifying units of width and precision in std::format cplusplus/papers#618

[tahonermann]

SG16 plans to discuss this comment Tuesday morning:

• http://wiki.edg.com/bin/view/Wg21belfast/SG16

[JeffGarland]

Reviewed Monday in LWG small group.

[tahonermann]

SG16 in Belfast: Accept US228 with P1868R0 modified to specify the
encoding used to interpret the input text for the purposes of width and
precision computation be implementation defined, but not locale
sensitive, and with non-normative guidance to use Unicode if possible.

Rationale: Victor's demonstration that the algorithm described in
P1686R0 produced improvements over the classical printf behavior for
common and intended use cases was convincing. This resolution also
reflects existing practice as exhibited by the POSIX wcswidth() function.

This approach had strong consensus (11 in favor, 1 neutral, no against,
no abstentions). Consensus was reached by polling options for various
aspects of the P1868R0 proposal:

• that field units should correspond to display width.
• the method of width and precision computation (based on the Unicode
  code point ranges specified in P1868R0 vs implementation defined
  behavior).
• the encoding used to interpret input text for the purposes of width
  and precision calculation (implementation defined but not locale
  sensitive with guidance for Unicode vs execution character set vs
  native character set (locale dependent) vs a Unicode encoding).

We did not have consensus for pursuing an opt-in specifier such as 'L'
for use of the native (locale sensitive) character set.

Removing the SG16 label. This should next be reviewed by LEWG.

SG16 minutes are available at:

• http://wiki.edg.com/bin/view/Wg21belfast/SG16NBUS228

[tituswinters]

LEWG in Belfast: Forward D1868R1 to LWG for C++20.

(Prioritization on this vs. other paper-sized NB comments will be shared with LWG chair when LEWG finishes NB comment processing.)

[JeffGarland]

LWG Tuesday afternoon in Prague - adopt for c++20.

[jensmaurer]

Accepted with modification. See paper P1868R2.