How to make a HTTP request synchronously in a coroutine

[rsp4jack]

I have tried this code:

require("coroutine")

local imdone = false

local function threadhttp()
    local request = http.get("https://example.tweaked.cc")
    print(request.readAll()) -- (2)
    request.close()
    imdone = true
end

local coro = coroutine.create(threadhttp)

while imdone == false do
    coroutine.resume(coro)
    print(coroutine.status(coro))
    sleep(0.1)
end

Output:

suspended
suspended
...(repeat)
(Ctrl+T)

HTTP server has already responded but code will never be executed to (2).
How to make a HTTP request synchronously in a coroutine?

CC:T Version: cc-tweaked-1.18.2-1.101.0

[SquidDev]

Aaaah, that's not quite how coroutines and the event system work. You need to resume the coroutine with events you've pulled from further up the coroutine tree. See this recent discussion for a example of how you should be resuming coroutines!

[SquidDev]

Though in this case, you could probably get away without using http.request` and the events directly instead.

[Lupus590]

Adding to this, you should avoid using sleep as it will eat events that are not its timer event. Use os.pullEvent, os.pulleventRaw or coroutine.yield instead.

[rsp4jack]

Now it works:

require("coroutine")
local pretty = require "cc.pretty"

local imdone = false

local function threadhttp()
    local request = http.get("https://example.tweaked.cc")
    print(request.readAll())
    request.close()
end

local coro = coroutine.create(threadhttp)

local ok, filter = coroutine.resume(coro)
while coroutine.status(coro) ~= "dead" do
pretty.pretty_print(filter)
  local event = table.pack(os.pullEvent(filter))
  pretty.pretty_print(event)
  ok, filter = coroutine.resume(coro, table.unpack(event, 1, event.n))
end
if not ok then error(filter, 0) end

Thanks for your explanation!

[rsp4jack]

BTW, it is confusing to a beginner. I think there should be some guides about coroutine and event system in docs.

[SquidDev]

BB's guide to coroutines is probably the best introduction to the whole thing.

The thing is coroutines are pretty niche. 90% of the time you don't need to worry about coroutines, and when you do, almost all of the time you can just get away with using the parallel API, so there's definitely more important topics we should document first.