error_code parameter not reset in directory_entry::status()

[gfgtdf]

The documentation https://www.boost.org/doc/libs/1_83_0/libs/filesystem/doc/reference.html#status specifies

If, during attribute determination, the underlying file system API reports an error, sets ec to indicate the specific error reported. Otherwise, ec.clear().

But it was observed that ec.clear() is no longer called in all cases, see for example wesnoth/wesnoth#7866

[Lastique]

Please, provide a reproducer or at least more information as to what's going on. ec.clear() is being called here.

[gfgtdf]

I see i linked the wrong documentation, this issue is about directory_entry::status() not the status() free function, I think what i said still applies though: https://www.boost.org/doc/libs/1_83_0/libs/filesystem/doc/reference.html#directory_entry-observers refers to https://www.boost.org/doc/libs/1_83_0/libs/filesystem/doc/reference.html#Error-reporting which states:

Functions having an argument of type system::error_code& report errors as follows, unless otherwise specified:

If a call by the implementation to an operating system or other underlying API results in an error that prevents the function from meeting its specifications, the system::error_code& argument is set as appropriate for the specific error. Otherwise, clear() is called on the system::error_code& argument.

Here

filesystem/include/boost/filesystem/directory.hpp

Line 192 in e65ddb6

if (!filesystem::status_known(m_status))

ec is not reset when the status is already known, which is a behavior change from boost 1.82

[Lastique]

Oh, sorry, you mean directory_entry members. Those are not documented to call ec.clear(), but in std::filesystem they are effectively required to clear the error code.

I'll fix this, thanks for the report.

[gfgtdf]

Thanks for fixing this.