bug#40407: [PATCH] slow ENCODE_FILE and DECODE_FILE

[Eli Zaretskii]

> From: handa <handa@gnu.org>
> Cc: hirofumi@mail.parknet.co.jp, mattiase@acm.org, 40407@debbugs.gnu.org,
>  handa@gnu.org
> Date: Thu, 16 Apr 2020 22:11:37 +0900
> 
> > I prefer to wait for Handa-san's response, and meanwhile install the
> > least disruptive change, which just fixes the one aspect that got
> > broken.  Call me a coward, if you wish.
> 
> I think Mattias' patch is good.

Including making the coding-type of utf-7 'charset'?  I think that
didn't work, see an earlier message in this discussion:

  https://debbugs.gnu.org/cgi/bugreport.cgi?bug=40407#107

And could you please tell more about the conditions for a
coding-system to be a candidate for 'charset' coding-type? what
exactly is such a coding-system supposed to do/provide/support?  I
don't think this is documented anywhere, and most/all of the
coding-systems that have this type are simple single-byte encodings
that support just 256 codepoints (which is not what utf-7 is).

Thanks.