Golang like channels

[vulpesx]

@dee0xeed

In essence, coroutines are non-local gotos (aka longjumps in C stdlib).
And goto statements are “considered harmful”, aren’t they?

goto is necessary, in some form. function calls a re goto, any branch is goto.

goto is bad when you can do it arbitrarily, especially when the language doesnt provide a less arbitrary goto for whatever purpose you need.

zig has plenty of goto even a somewhat arbitrary one, labelled switch, but its still limited in its own scope and is clearly labelled.
zig doesnt provide a goto as arbitrary as C’s goto or longjump. If you truly need that level of goto your only option in zig is assembly.

@dee0xeed

Co-routine is a generalization of sub-routine
and I would not say it’s easier to learn.
From a view-point of a user (programmer) - may be…
From a view-point of a developer of a language - hardly.

I was talking from the perspective of a programmer, developers of languages implement coroutines because programmers want them, because they have used them and like them, because they are relatively easy to learn to use.

[awesomo4000]

Here is another way to do coroutines (the way Erlang does it) through automated source to source translation:

GitHub - kjellwinblad/yielding_c_fun: Source-to-source transformer that creates...

[dee0xeed]

@vulpesx

function calls are goto, any branch is goto

These two are a bit different on CPU side

• calling a subroutine requires (at least) storing an IP (“instruction pointer”) / PC (“program counter”) (on stack) to know where to return to.
• jmp / br (and plenty of conditional variants) instructions do not.

@vulpesx

zig has plenty of goto even a somewhat arbitrary one, labelled switch, but its still limited in its own scope and is clearly labelled

Any “systems programming” language can’t do without
keywords, more or less directly compiled into jmp/br
(continue, break, explicit goto and alike)…

… well… even bash has them (continue and break),
but it’s definitely not “systems programming” language, rite?

@vulpesx

but its still limited in its own scope and is clearly labelled

yes, in C you can jump into a for loop from a code before that loop,
as well as from a code after ttah loop,
but show me who is doing such nuts ever?

@vulpesx

zig doesnt provide a goto as arbitrary as C’s goto or longjump

• Zig has just replaced well-known C’s cleanup pattern (goto __cleanup) with more elaborated defer/errdefer pattern (yes, it’s really a sort of innovation)
• setjump/longjump were/are just forerunners for numerous implementations of async/await in various languages

once again

• jumps / gotos within a function/subroutine (when used “properly”) are just fine
• non-local jumps (yields) are much more harder to deal with in more or less complex code than… than event-driven-state-machines (imho)

[nubskr]

hi, a bit new to the party here, but I built something similar: GitHub - nubskr/ziggy: High performance lock free MPMC channel in under 200 LOC

performance competitive with crossbeam channel for the most part