On 17/09/2019 10:46, Lassi Kortela wrote:

>
> I always start from a binary protocol, and ask whether a textual one
> saves anything. It's so much easier to avoid all the pitfalls.
>

A tangential but related issue is that I shy away from using read and
write for "protocols" where possible due to the frequent presence of
vendor extensions that allow for executing arbitrary code in read -
which isn't a problem in trusted environments, but still seems like an
unnecessary danger that could lead to an exploit escalating... I've
ended up writing ad-hoc binary representations for protocol use when it
would have been far nicer to use read and write. An untrusted-data-safe
binary sexpression format SRFI would be most welcome with me - but I'd
like it in its own SRFI rather than a persistence one, as I'd use it for
protocols more than for persisting :-)

--
Alaric Snell-Pym   (M7KIT)
http://www.snell-pym.org.uk/alaric/