telegraf: Awaiting for telegraf.launch() stops the code.

Please update the documentation. Spent quite a lot of time before found out that promise is never resolved by design…

You could also do this:

bot.botInfo = await bot.telegram.getMe();
console.log("Bot started");
bot.launch();

This will prevent launch() from calling getMe on its own, and has all the same guarantees pre-4.11.0 had.

Thanks for opening this issue.

Since 4.11 (4.11.0 release notes: see “Other changes”), we made a choice to make bot.launch() throw error if long polling errors (related: #1657). This would unfortunately mean that the promise never resolves until the bot stops.

We’re aware that although the previous behaviour was undocumented, many users have relied on it to get a “bot started” event. I’m holding off on “fixing” this prematurely by rolling back, so that we can still provide a way for errors to be caught. Meanwhile, you can do this:

const config = require("./config.json");
const { Telegraf } = require(https://github.com/telegraf/telegraf);

(async () => {
    const telegraf = new Telegraf(config.telegram.token);
    console.log("Telegram started.");
    await telegraf.launch();
})();

Note that awaiting bot.launch() is not strictly required unless you have a usecase where you need to catch errors. See the release notes for more details.

I’ll leave this issue open until we’ve provided a new solution to both sides.

Here we go! #1949 adds a callback to bot.launch that is called when launching the bot.

bot.launch(() => console.log("Bot is starting!"));

If you pass LaunchOptions, the callback goes after the options.

bot.launch({ dropPendingUpdates: true }, () => console.log("Bot is starting!"));

This is available on npm as a release candidate: 4.16.0-rc.2. Stable release when #1949 is merged.

Thanks! It works, but as you said, I won’t close the problem until you come up with a better solution.

I solved the problem, for me this solution is suitable. Telegraf@4.14.0

const bot = new Telegraf(process.env.TOKEN);
bot.launch();
// Started message
if (bot) bot.telegram.getMe().then((res) => console.log(`Bot started on https://t.me/${res.username}`))

I hope, I solved your question.

> npm install telegraf@4.10.0

added 1 package, changed 1 package, and audited 109 packages in 2s

11 packages are looking for funding
  run `npm fund` for details

found 0 vulnerabilities
> node .\a.js         
  telegraf:main Created a `Telegraf` instance +0ms
  telegraf:main Connecting to Telegram +2ms
  telegraf:client HTTP call getMe {} +0ms
  telegraf:main Launching @<botname>_bot +460ms
  telegraf:client HTTP call deleteWebhook { drop_pending_updates: undefined } +460ms
  telegraf:polling Starting long polling +0ms
  telegraf:client HTTP call getUpdates { timeout: 50, offset: 0, allowed_updates: [] } +89ms
  telegraf:main Bot started with long polling +91ms
Telegram started.
> 

@mindrunner

How does this solve the original issue? launch still blocks forever when being awaited which leaves a dangling promise.

bot.launch returning a long-lived promise is intentional, because it’s the only place you can handle polling errors (which are generally fatal and the bot cannot continue, unlike handler errors). It allows #1657 to recover from errors gracefully, for when there are more than the single bot running in the process. The new launch callback solves the problem of wanting to log a message on bot start.

Read https://github.com/telegraf/telegraf/releases/tag/v4.11.0 for more details on why that change was made.

I solved the problem, for me this solution is suitable. Telegraf@4.14.0

const bot = new Telegraf(process.env.TOKEN);
bot.launch();
// Started message
if (bot) bot.telegram.getMe().then((res) => console.log(`Bot started on https://t.me/${res.username}`))

I hope, I solved your question.

I don’t think this works, my test failed at telegraf@4.14.0. I forced the telegraf version to 4.10.0 after read the changelog. I will consider upgrading when this issue closed.