Use `get_network()` instead of global `$current_site`

[flixos90]

get_network() falls back to the current network if no parameter is provided. That means we can use it anywhere we currently work with globals or with get_current_site(). I think the latter function should be deprecated afterwards.

[paulwilde]

See [38388].

[flixos90]

@wonderboymusic: Seeing the changeset above, not sure if that's the right step as we should start to use get_network() instead of get_current_site() or the global $current_site. Maybe we should make the new functions get_site() and get_network() compatible with non-multisite and then move these too.

[flixos90]

In 37414.diff​ I replaced all instances of get_current_site() with get_network(), plus some related adjustments, like changing internal variable names from $current_site to $current_network.

While get_network() is not available on some of these functions, this should not be a problem since there are is_multisite() checks in place that prevent it from being called. This was similar with get_current_site() as well, so I was not sure why the function was moved in 38388. The global $current_site is not available on non-multisite, so I think that function should remain in an ms-*.php file.

[jeremyfelt]

[38388] has been reverted in [38636]. Due to this and possibly some other churn, 37414.diff​ is having some trouble applying cleanly.

I like the idea, especially from a naming perspective. This change would also provide the filter in places where it wasn't available before, which is probably good. It does mean that apply_filters() fires that many more times in a page view, though I have not studied to see if there's a real enough performance impact because of that.

It could be good to do some profiling, but this seems like a nice change.

[flixos90]

Replying to jeremyfelt:

[38388] has been reverted in [38636]. Due to this and possibly some other churn, 37414.diff​ is having some trouble applying cleanly.

Let's see if we can get #34450 in first as this would contain another case that this ticket would apply to. I can take care of creating an updated patch afterwards.

[jeremyfelt]

Replying to flixos90:

Replying to jeremyfelt:

[38388] has been reverted in [38636]. Due to this and possibly some other churn, 37414.diff​ is having some trouble applying cleanly.

Let's see if we can get #34450 in first as this would contain another case that this ticket would apply to. I can take care of creating an updated patch afterwards.

And #34450 is in. :)

[flixos90]

37414.2.diff​ is a new patch that works against current trunk. Other than that, it does the same as mentioned above in https://core.trac.wordpress.org/ticket/37414#comment:4.

A site note: While doing the replacements, I found two minor issues that are a bit related, #38246 and #38247. Maybe those should be merged before this one.

[jeremyfelt]

37414.3.diff​ is another refresh.

Things are looking good.

• I did some brief performance testing between the global $current_site() and get_network() and things are only slightly slower. In most cases, this will be pulling from something that's already available. The only real overhead is the application of the get_network filter on every usage, which is a nice side effect on its own.
• All unit tests in single and multisite are passing. We aren't using get_network() or get_current_network_id() in any place that is not multisite.
• Single site install, multisite setup, multisite management, new site signup, are all working as expected. This covers some of the things that are kind of hard to cover in our tests.

Let's do it. :)

[jeremyfelt]

In 38814:

Multisite: Use get_network() and get_current_network_id() for current network data.

get_network() falls back to the current network when called without any arguments. Between this and get_current_network_id(), we can replace almost all instances of the global $current_site and all instances of get_current_site().

This effectively deprecates get_current_site(), something that we'll do in a future ticket.

Props flixos90.
Fixes #37414.