# Relaunch the task bound to channel

**URL:** <https://discourse.julialang.org/t/relaunch-the-task-bound-to-channel/60777>\
**Category:** General Usage\
**Created:** [May 8, 2021, 1:50pm UTC](https://discourse.julialang.org/t/relaunch-the-task-bound-to-channel/60777 "2021-05-08T13:50:46Z")\
**Posts on this page:** 3\
**Page:** 1

<div class="post-metadata">

**Author:** ![zekeriya.sari](https://sea2.discourse-cdn.com/julialang/user_avatar/discourse.julialang.org/zekeriya.sari/32/13695_2.png) [@zekeriya.sari](https://discourse.julialang.org/u/zekeriya.sari)\
**Post date:** [May 8, 2021, 1:50pm UTC](https://discourse.julialang.org/t/relaunch-the-task-bound-to-channel/60777/1 "2021-05-08T13:50:46Z")

</div>

Is it possible to relaunch a task bound to a channel so that the data can be recollected from the channel?. Here is a MWE:

```julia
julia> ch = Channel(0) do ch 
       for i in 1 : 5 
       put!(ch, i)
       end 
       end
Channel{Any}(0) (1 item available)

julia> collect(ch)
5-element Vector{Any}:
 1
 2
 3
 4
 5

julia> collect(ch) 
Any[]

```

To do this, I tried to re-initialize the task bound to the channel, but I got an exception saying that the channel is closed.

```julia
ulia> ch = Channel(0) do ch 
       for i in 1 : 5 
       put!(ch, i)
       end 
       end
Channel{Any}(0) (1 item available)

julia> f = ch.cond_put.waitq.head.code
#517 (generic function with 1 method)

julia> collect(ch) 
5-element Vector{Any}:
 1
 2
 3
 4
 5

julia> collect(ch) 
Any[]

julia> @async f() 
Task (failed) @0x00007f47aaa60160
InvalidStateException("Channel is closed.", :closed)
Stacktrace:
 [1] check_channel_state
   @ ./channels.jl:170 [inlined]
 [2] put!
   @ ./channels.jl:314 [inlined]
 [3] (::var"#13#14")(ch::Channel{Any})
   @ Main ./REPL[5]:3
 [4] (::Base.var"#517#518"{var"#13#14", Channel{Any}})()
   @ Base ./channels.jl:132
 [5] (::var"#15#16")()
   @ Main ./task.jl:406

```

---

<div class="post-metadata">

**Author:** ![pbayer](https://sea2.discourse-cdn.com/julialang/user_avatar/discourse.julialang.org/pbayer/32/11675_2.png) [@pbayer](https://discourse.julialang.org/u/pbayer)\
**Post date:** [May 9, 2021, 2:02pm UTC](https://discourse.julialang.org/t/relaunch-the-task-bound-to-channel/60777/2 "2021-05-09T14:02:04Z")

</div>

There are almost two problems with that approach:

- after the `ch = Channel(0) do ch for i in 1 : 5 ...` loop the task has finished and
- with the `collect(ch)` the channel is closed.

If you want repeatedly receive data from a channel, you need

1. a persistent server task and
2. for the client/server interaction you need some convention, for example
  - a messaging pattern or
  - if the server delivers continuously on the client side some `collect_data` function, where you can specify the number of items to collect.

A messaging pattern could follow the following lines:

```julia
# define the messages
struct Request
    n::Int
end

struct Response{T}
    y::T
end

struct Stop end

# define the server behavior
function serve(req, res)
    i = 1
    while true
        msg = take!(req)
        if msg isa Request  
            put!(res, Response(collect(i:i+msg.n-1))); 
            i = msg.n+1
        elseif msg isa Stop 
            break
        end
    end
end

# define the channels
a = Channel()
b = Channel()
# start the server
t = Threads.@spawn serve(a, b)

```

Then you can do

```julia
julia> put!(a, Request(5))
Request(5)

julia> take!(b).y
5-element Vector{Int64}:
 1
 2
 3
 4
 5

julia> put!(a, Request(5))
Request(5)

julia> take!(b).y
5-element Vector{Int64}:
  6
  7
  8
  9
 10

```

Then you can continue to request/receive data as long as you want and stop the server task at some point.

---

<div class="post-metadata">

**Author:** ![zekeriya.sari](https://sea2.discourse-cdn.com/julialang/user_avatar/discourse.julialang.org/zekeriya.sari/32/13695_2.png) [@zekeriya.sari](https://discourse.julialang.org/u/zekeriya.sari)\
**Post date:** [May 10, 2021, 7:46pm UTC](https://discourse.julialang.org/t/relaunch-the-task-bound-to-channel/60777/3 "2021-05-10T19:46:29Z")

</div>

Thank you @pbayer for your detailed reply.

But the problem is not to _repeatedly_ receive data from a channel. In such a case, as you did in your example, producer/consumer or server/client tasks can be created and bound to channels such that the data flow from server to the client be provided until the server’s task is poison-pilled (with Stop in the example).

But, the problem is exactly what I explained in my MWE. _After_ a channel is closed and the task bound to the channel is done, can we somehow _reopen_ the channel and _relaunch_ the task so that the data flow from the client to server starts again?