Re: [boost] [review] Review of Nowide (Unicode) starts today

Subject: Re: [boost] [review] Review of Nowide (Unicode) starts today
From: Zach Laine (whatwasthataddress_at_[hidden])
Date: 2017-06-19 17:43:34

On Mon, Jun 19, 2017 at 12:37 PM, Frédéric Bron via Boost <
boost_at_[hidden]> wrote:

> Question to all:
>
> Why should we try to handle wrong UTF-16 (or wrong UTF-8)?
> 1. such files should not exist
> 2. if they exist, why?
>   - if it is because it is an old file, can the user just rename it
> properly?
>   - if it is because it was produced by a program, why should this
> program continue to work without fixing? Isn't it the best way that we
> get wrong filenames forever?

This came up at C++Now last month.  My understanding, from talking to
people who seem to know about this, is that such filenames are considered
valid by Windows.  To issue an error in such a case means not allowing
users to access files that Windows sees as well-formed.

Zach