bug#20231: 25.0.50; package.el mangles non-ascii content in downloaded archive-contents data

[Eli Zaretskii]

> From: Ivan Shmakov <ivan@siamics.net>
> Date: Sat, 11 Apr 2015 19:30:21 +0000
> 
> >>>>> Eli Zaretskii <eliz@gnu.org> writes:
> 
> […]
> 
>  >> As insert-file-contents may imply a whole lot of additional
>  >> processing beyond decoding UTF-8, I’d rather use an explicit
>  >> decode-coding-region here
> 
>  > Can you explain why?  What additional processing can get in the way?
> 
>       One of the examples I can readily think of is enriched-decode,
>       which will automatically be called should there be a
>       ‘Content-Type: text/enriched’ line near the top of the file
>       being read (per format-alist.)

Yes, but is that kind of processing harmful for the case at point?

Also, are we sure we will never get a compressed file, for example, or
some other binary file?  In those cases, decoding is not TRT.

>       I have no example at hand of how such processing may be
>       triggered in practice, but avoiding it altogether seems to me
>       like a saner approach.

It is also slower.  Again, I have no idea if speed matters in this
case.