2816. resize_file has impossible postcondition

Section: 31.12.13.34 [fs.op.resize.file] Status: C++20 Submitter: Richard Smith Opened: 2016-11-07 Last modified: 2021-06-06

Priority: 3

View all issues with C++20 status.

Discussion:

resize_file has this postcondition (after resolving late comment 42, see P0489R0):

Postcondition: file_size(p) == new_size.

This is impossible for an implementation to satisfy, due to the possibility of file system races. This is not actually a postcondition; rather, it is an effect that need no longer hold when the function returns.

[Issues Telecon 16-Dec-2016]

Priority 3

[2018-01-16, Jonathan provides wording]

[2018-1-26 issues processing telecon]

Status to 'Tentatively Ready'

[2018-3-17 Adopted in Jacksonville]

Proposed resolution:

This wording is relative to N4713.

[Drafting note: I considered a slightly more verbose form: "Causes the size in bytes of the file p resolves to, as determined by file_size ( [fs.op.file_size]), to be equal to new_size, as if by POSIX truncate." but I don't think it's an improvement. The intent of the proposed wording is that if either file_size(p) or truncate(p.c_str()) would fail then an error occurs, but no call to file_size is required, and file system races might change the size before any such call does occur.]

  1. Modify [fs.op.resize_file] as indicated:

    void resize_file(const path& p, uintmax_t new_size);
void resize_file(const path& p, uintmax_t new_size, error_code& ec) noexcept;

    -1- Postconditions: file_size(p) == new_sizeEffects: Causes the size that would be returned by file_size(p) to be equal to new_size, as if by POSIX truncate.

    -2- Throws: As specified in 31.12.5 [fs.err.report].

    -3- Remarks: Achieves its postconditions as if by POSIX truncate().