Le 28/05/2020 à 02:29, Steve Litt a écrit :
> My experimentations with Guile
> ( http://troubleshooters.com/codecorn/scheme_guile/index.htm ) produced
> no results indicating any kind of intermittent or unexpected behavior
> in Guile.
>
> Guile is pure functional programming: With a few eceptions (like
> printing), there is no state and no side effects. Loops are done with
> recursion, best done with tail recursion. There is a purity of function
> unavailable from OOP bolt-ons like C++, Perl, and to a lesser extent
> Python. Anyone understanding recursion, functional programming and
> lambdas can handle Guile, at least for reasonably simple code. As long
> as the Guile interpreter is available on a mounted drive in early boot,
> I see no reason for caution about Guile.
>
> I think 80% of us grew up with Procedural or OOP languages and are
> familiar with them. Just like OOP requires different thought patterns
> than procedural, functional languages require (much) different thought
> patterns than Procedural or OOP. So Guile might prove a challenge to
> some, not because it's defective or complex in any way, but because
> functional programming requires much different thought patterns.

    Computer languages are IMHO the best concentrate of human
intelligence which have been put into the domain of software. The closer
they are to human concepts, the most power and freedom they give to the
programmer. In this respect, functionnal languages are certainly more
expressive than imperative ones.

    However Shepherd as PID1 goes against the common wisdom that init
must be oversimple - see the exemple of Rich Felker, but Laurent Bercot
would certainly be on the same side. Shepherd might live with sysv-init,
in particular if it could be a subreaper.

        Didier