Fix copy/paste-ability of default URL presented on startup

[ghost]

Currently the default URL message given on the console on startup is:

Copy/paste this URL into your browser when you connect for the first time,
to login with a token:
http://(myip.com or 127.0.0.1):8888/?token=8fdc8 ...

This will always need editing to use. Replace brackets/ 'or' part with host IP (e.g. 'myip.com')
to make it copy/pastable again.

[minrk]

This is the same as #3356, which was reverted because it caused problems. See #3605, #3668, #3703 for why we don't do this, and various discussions about what to do and when.

The latest proposal in #3947 is to display both the hostname and localhost URL in copy-pasteable form, which might be the best so far. Would you like to take a stab at doing that?

[ghost]

Proposal stabbed at, comments welcome.

[Carreau]

Thanks,

Your editor seem to be configured to trim whitespaces, this adds a lot of noise in the diff, and also often is the source of issues when trying to backport Pull-requests / use Git blame, or git-bisect.

We in general are in favor of removing whitespace, but only in proximity of modified code.

How comfortable are you in rebasing without the whitespace changes ? Otherwise we can do that for you.

[ghost]

Hi, thanks for looking at this PR.

The whitespace changes were in a separate commit for the reasons you quote, but I've just reverted that anyway.

I've also rebased my local branch to the head of master and force-pushed, which I assume was causing the previous merge error. There's been some issues with github's UI/backend whilst this went on, so please let me know if anything looks untoward.

Thanks,

Mark

[Carreau]

The whitespace changes were in a separate commit for the reasons you quote, but I've just reverted that anyway.

Actually as the last two commits were the whitespace and faulty one I've hard reseted to HEAD~2 so they are just not there.

I'll wait until github full restoration of service to merge.

[ghost]

The CI pipeline looks to be stuck on this PR, given that the same code changes passed a few days ago. Can it be restarted?

[ghost]

Hi @Carreau, just checking in - is there anything I can do to help merge this PR? I don't believe the current CI checks are sane.

[takluyver]

@Einon generally closing and reopening a pull request will re-run the CI. I'll try that now.

[takluyver]

We've previously been trying to show only the real IP if there's one set, and make the 'or 127.0.0.1' a special case when you're listening on all IPs. This would show the 127.0.0.1 option all the time, even when you're not listening on that IP, and even when it's the same as the first address. But maybe that's better than trying to be clever about it?

[ghost]

Hi @takluyver, thanks for the update.
In my experience it would be an unusual system not to be listening on 127.0.0.1, the loopback address. Is that an existing use case for jupyter notebooks?

[takluyver]

It's definitely possible to listen only on an external-facing interface, though I don't know how many people do it. But even when you do that, it might be e.g. the external network interface of a container, which is mapped to the same port on localhost outside the container. There's no way in general to know what address the server is actually accessible at.

So I'm inclined to merge this, and we can always reconsider it again later.

[AlJohri]

Thanks for this! I was just about to make an issue regarding the copy/pastability.

[Markus92]

Would it be possible to apply this patch to the 5.7.x branch? It'd allow me to put it into production right away (and make our users very happy!).