Deno Fetches, Node Doesn’t; Why I Had to Put Node down

Photo by Jamie Street on Unsplash

I spent the last 2 days of my life battling Node and Webpack.

There is some weird import/require module thing that is causing the code I’m writing that depends upon the js-ipfs / ipfs-http-client module to break with an error like:

TypeError: globals.fetch is not a function

As I tracked it down it appears this is due to webpack in ‘node’ mode.

When I run the same code, excluding webpack, it runs fine. This code also previously ran fine in the browser.

After two days of frustration and research I’ve come to the unhappy conclusion that here is a sort of endless circle of everyone wanting to pollyfill fetch, and the webpack ‘node’ mode not knowing how to deal with it sanely.

The Pain of Teaching and Old Dog a New Trick

Node does not support fetch. See this github issue for all the ugliness of why.

This by Benjamin Gruenbaum sums it up nicely

To be clear fetch is currently only blocked in core because it's a lot of work and effort to add to core well and there are a lot of stepping stones first (like figuring out what to do with AbortController and WHATWG streams). Fetch is not blocked in core because we don't want it in core.

Deno Fetches

It turned out that the solution to my problem was to give up on node and move on to Deno who does know how to fetch.

To use Deno as a successful Node replacement in my vscode environment all I needed to do so far was make the typings available via:

$ deno types > src/types/deno.d.ts

After that my fetch problem went away and my code ran fine.

I also no longer needed to have webpack in ‘node’ mode, which feels even better.

Note that I’m using deno via feeding it a single main.js file ‘transpiled’ using webpack. At the moment, I’m not interested in Deno’s take on ‘import urls’, and so prefer to use the webpack/npm tooling.

Deno appears to be OK with this state of affairs, and just like that I became a Deno supporter.

What to do with Old Dogs?

Given that fetch isn’t IMHO that difficult of an API to support, if one starts from the proper primitives, and Node is clearly struggling with this critical API…I think this signals that Node is perhaps better thought of as a legacy tool.

I personally would not invest any further effort in retrofitting this venerable work-horse with more modern concepts or APIs. Nor would I invest time targeting it with new code. At some point it is better to let the past be, and start building fresh with lessons learned.

Perhaps Deno will be a replacement. Better IMHO would be for several replacements to come to the forefront.

For now I will start training my new (hopefully) faithful companion, and will give the old dog a much deserved rest.

Rest well old friend.

Originally published at https://github.com.