feat: reduce bundle size by 1/3

[mikeal]

I did a few things here;

1. replaced the Python script that generated 3 files with a Node.js script that generates only one file.
   • the module it outputs is also a bit different in order to reduce the size of the file more, even after minification/compression.
2. replaces 2 codegen files with files that create their tables from the other generated file. this means a much smaller bundle and a negligible increase in import time.
3. updates aegir and fixes new lint issues. i would have made this another PR but the version that was here was before aegir lint —fix was implemented so in order to get the linter passing I need to upgrade it.

[mikeal]

the build is mad about the commit message, are we actually trying to be this strict about commit messages?

[vmx]

are we actually trying to be this strict about commit messages?

I think especially this case makes sense as there's tooling around the changelog, so it makes sense to have a fixed set of categories.

[vmx]

I'd like to get #51 merged before this PR as I added a test that fails with this PR :)

[hacdias]

@mikeal can you rebase and check the build with the new tests, please? :)

[mikeal]

My plate is pretty full for the next month or so, maybe @vmx can take this over if he has some cycles.

[hacdias]

I can take a look at it myself!

[hacdias]

@vmx I fixed the test by only the adding the first element to the print table. In other words, since ipfs appears after p2p on the original table, it was getting overridden. The issue with this is: order matters. But I guess it already mattered on master.

[vmx]

The issue with this is: order matters. But I guess it already mattered on master.

Yes it was like this in master. The important part here is that we prefer p2p over ipfs, which was the reason why I added the test.

[hacdias]

@vmx just went through the code and made some simplifications. Also, accepted your suggestion of saving the table to a .json file instead!

[vmx]

@hacdias Nice! I actually haven't thought of actually storing it as a JSON file. I just thought, using JSON somehow might be an idea, but that's now even better!

@mikeal As things changes quite a bit from your code, would you mind having another quick look over this PR (or please leave a comment if you can't be bothered :)

[alanshaw]

Anything blocking this from getting merged and released?

[hacdias]

@alanshaw we were just waiting for @mikeal review. Other than that, nothing! Wdyt?

[alanshaw]

@hacdias I think that this is soon to be the biggest module in ipfs-http-client and I'd like to reduce it by 1/3.

As the lead maintainer of this module you should feel empowered to take ownership of this PR and merge it if you see fit 😁

If it were me, I'd double check the tests are passing, that the size reduction is still as anticipated and link it in to js-ipfs and run the tests to ensure nothing broke.