Building an updatable Slack message
Slack is fully awesome. At wehkamp we use it for our internal communication and as a tool for our DevOps. The Slack API allows us to build even more advanced integrations. In this blog I’ll explore how to use the API to create stuff like a powerful progress indicator, just by updating a Slack message:
We use Axios as a promise-based HTTP client to connect to Slack. Each call needs an authentication-token and a channel ID (even for private messages). There are two end-points we are going to be using:
- chat.postMessage — this will send a new message to the channel and give us a message identifier
ts. We will store and reuse it to update the message.
- chat.update — this will update the message that corresponds to the
Let’s create a function that uses both end-points to send a message:
The function returns the message identifier in the form of a promise. The only thing we need to do is save the message identifier and reuse it when calling the function.
The first problem I ran into had to do with messages not arriving in the same order that I expected. Sometimes I ended up with an earlier message as end-result. And sometimes I wanted to send an update for a message that had not even been created yet. If we want to work asynchronously we need to synchronize the way we send messages.
We need to build a class that will handle the process of sending messages. It will store the message identifier so we can update the message. The class should also make sure that only a single message is being send at the same time.
If a message is being send and a new message comes in, the class should store that new message and send it after the first message has been sent.
But what if a 3rd message comes in? We are only interested in sending the latest message, so we should only store the last message. A pattern like this prevents useless updates / excessive use of the Slack API. Be aware: rate limiting applies to the Slack API, so you might want to consider only updating the message once a second.
So let’s look at the implementation:
By implementing a simple Boolean that checks if a message is being send, we solve most — if not all — of our asynchronous problems.
This class is used in the bot-zero project to show an example of a progress indicator. Go check it out and let us know what you think.
Originally published at keestalkstech.com on October 11, 2018.