Re: libnettle/libhogweed WIP

[Eli Zaretskii <eliz@gnu.org>]

> From: Stefan Monnier <monnier@iro.umontreal.ca>
> Date: Wed, 19 Apr 2017 23:36:11 -0400
> 
> >> In my experience it's important for the coder to carefully choose which
> >> coding-system should be used.  So if we want the functions to do
> >> encoding internally, then I'd prefer we have a non-optional
> >> coding-system argument.  "Automatically choose an appropriate
> >> coding-system" encourages bugs.
> > How is this different from write-region and its ilks?
> 
> Off the top of my head:
> - it's new, so we get to avoid past errors

Not sure what past errors you had in mind.  Any errors we made in the
encoding/decoding department were fixed by Emacs 23, and the stuff is
remarkably stable since then, with a single minor improvement in Emacs
24.4.  From my POV, this one of the greatest success stories in Emacs.
Paid for with sweat, blood, and tears, but success nonetheless.  Why
would we want to refrain from reusing it?

> - it's not a command

I don't see how this is relevant: write-region is mostly used
non-interactively.

> - it's not limited to interaction with files

write-region was just an example; we use basically the same rules with
process I/O.