feat: ipns publish example

[achingbrain]

This follows on from @DougAnderson444's great work in #3166 (I can't push commits to his fork).

I've tided up the test and the instructions, but it doesn't always work reliably for me.

When I hit Publish to IPNS, quite frequently the IPNS name is resolved to the empty directory CID:

// browser console
Publish to IPNS
Initial Resolve QmPnx5euUoR7mHU3nf29tcRn5yCb1X7r6QHRTS9RpT13Wf
Subscribing to /record/L2lwbnMvEiAVmG4UEkG8Obk4CpyfVwRhS14ou8rQbCiIZqWDKrtEAA
12:32:47 Message c6b2c0951cba22d8 from QmPnx5euUoR7mHU3nf29tcRn5yCb1X7r6QHRTS9RpT13Wf
#1) Topic: /record/L2lwbnMvEiAVmG4UEkG8Obk4CpyfVwRhS14ou8rQbCiIZqWDKrtEAA
Value: /ipfs/QmUNLLsPACCz1vLxQVkXqqLX5R1X345qqfHbsf67hvA3Nn
12:32:48 Message 5dc88986550867ea from QmPnx5euUoR7mHU3nf29tcRn5yCb1X7r6QHRTS9RpT13Wf
#2) Topic: /record/L2lwbnMvEiAVmG4UEkG8Obk4CpyfVwRhS14ou8rQbCiIZqWDKrtEAA
Value: /ipfs/QmUNLLsPACCz1vLxQVkXqqLX5R1X345qqfHbsf67hvA3Nn
Publishing /ipfs/Qmd938ZvbdDBaHGFY5QC2VFesRrj4wnszwpiT3pQ6vGati to self /ipns/QmPnx5euUoR7mHU3nf29tcRn5yCb1X7r6QHRTS9RpT13Wf
12:32:50 Message 3866650046515a58 from QmPnx5euUoR7mHU3nf29tcRn5yCb1X7r6QHRTS9RpT13Wf
#3) Topic: /record/L2lwbnMvEiAVmG4UEkG8Obk4CpyfVwRhS14ou8rQbCiIZqWDKrtEAA
Value: /ipfs/Qmd938ZvbdDBaHGFY5QC2VFesRrj4wnszwpiT3pQ6vGati
Published QmPnx5euUoR7mHU3nf29tcRn5yCb1X7r6QHRTS9RpT13Wf to /ipfs/Qmd938ZvbdDBaHGFY5QC2VFesRrj4wnszwpiT3pQ6vGati
Wait 5 seconds, then resolve...
Try resolve QmPnx5euUoR7mHU3nf29tcRn5yCb1X7r6QHRTS9RpT13Wf on server through API
Resolved: /ipfs/QmUNLLsPACCz1vLxQVkXqqLX5R1X345qqfHbsf67hvA3Nn
Error, resolve did not match /ipfs/QmUNLLsPACCz1vLxQVkXqqLX5R1X345qqfHbsf67hvA3Nn !== /ipfs/Qmd938ZvbdDBaHGFY5QC2VFesRrj4wnszwpiT3pQ6vGati

// server console
Subscribing to /record/L2lwbnMvEiAVmG4UEkG8Obk4CpyfVwRhS14ou8rQbCiIZqWDKrtEAA
12:32:47 Message c6b2c0951cba22d8 from QmPnx5euUoR7mHU3nf29tcRn5yCb1X7r6QHRTS9RpT13Wf
#1) Topic: /record/L2lwbnMvEiAVmG4UEkG8Obk4CpyfVwRhS14ou8rQbCiIZqWDKrtEAA
Value: /ipfs/QmUNLLsPACCz1vLxQVkXqqLX5R1X345qqfHbsf67hvA3Nn
12:32:48 Message 5dc88986550867ea from QmPnx5euUoR7mHU3nf29tcRn5yCb1X7r6QHRTS9RpT13Wf
#2) Topic: /record/L2lwbnMvEiAVmG4UEkG8Obk4CpyfVwRhS14ou8rQbCiIZqWDKrtEAA
Value: /ipfs/QmUNLLsPACCz1vLxQVkXqqLX5R1X345qqfHbsf67hvA3Nn
[Pass] Pubsub.ls
[Pass] Name.Pubsub.subs
12:32:50 Message 3866650046515a58 from QmPnx5euUoR7mHU3nf29tcRn5yCb1X7r6QHRTS9RpT13Wf
#3) Topic: /record/L2lwbnMvEiAVmG4UEkG8Obk4CpyfVwRhS14ou8rQbCiIZqWDKrtEAA
Value: /ipfs/Qmd938ZvbdDBaHGFY5QC2VFesRrj4wnszwpiT3pQ6vGati
Subscribing to /record/L2lwbnMvEiAVmG4UEkG8Obk4CpyfVwRhS14ou8rQbCiIZqWDKrtEAA

If I then press Publish to IPNS again, it works and the test completes (assuming it hasn't timed out by then).

@vasco-santos any ideas why you have to publish twice?

[DougAnderson444]

Thanks for the help with the test, it's helped me learn how to write them.

I noticed the empty directory CID too, could not figure out why it's happening. I originally thought it was related to missing {stream: false} but then I saw that @jacobheun found a bug in js http client streaming can't be overridden, which may be contributing to the issue?

[achingbrain]

No, generally you want streaming, it just means you get results as they come in rather than waiting for them all to arrive first.

This PR has that fix applied anyway so I don't think it's related.

[vasco-santos]

So, when IPFS is inited, we initializeKeyspace, which points to an empty directory:

• https://github.com/ipfs/js-ipfs/blob/master/packages/ipfs/src/core/components/init.js#L144

• js-ipfs/packages/ipfs/src/core/ipns/index.js

  Line 89 in 7aba835

  async initializeKeyspace (privKey, value) { // eslint-disable-line require-await

Taking this in mind, it seems that the first name.publish did not finish before the resolve, or somehow it is running before init finishes.