@test success
With this patch applied I could complete the installation of pkg_kunena_v3.0.6_2014-07-28.zip without issues.

+1

To whomever commits this, I'd also add a comment above the MooTools call and make sure Hathor is loading it in this view. A comment like this should do:

// MooTools is loaded for B/C for extensions generating JavaScript in their install scripts, this call will be removed at 4.0

@mbabker 4.0 or 3.5?

i tested Hathor, and kunena can install. So RTM

4.0, our B/C policy does not allow us to break B/C at any release but a major (X.0) release.

well... ok, although this is a very special case (we are not breaking the API), but again...

It technically isn't an API but it can be argued it's a B/C breaking change. It's one of those gray areas that you could argue every possible decision for weeks on end and get nowhere, at least this way a decision is being made and is clearly stated somewhere (even if it's buried in a code comment right now).

actually if we commit to keep it till 4.0, the comment is probably redundant: lots of things will break at 4.0, I suppose...

It's about the same as our @deprecated 4.0 tags in doc blocks. Redundant? Probably. Clear that we're pulling it out? Absolutely.

And also preventing that nobody will try to remove it before 4.0

Point taken!

I can live with it if that's the only place MooTools will still be loaded.
However it should be clear that this doesn't become a precedence for other places. Our goal since 3.0 is to get rid of MooTools. By now, every developer should know that it will be removed eventually and that they can't trust on it being present.

Don't we need it also for discover and update?

Try updating/discovering Kunena and you know :)

Thanks guys. We will look into removing last Mootools bits from our installer.

This is kind of special case as Kunena uses enqueued message to inject our installer code into the page and we have no control on what JS has been loaded to the page. :) In the future it would be great if installers had a bit more freedom to run their AJAX installers as Kunena upgrades can take minutes of time if there are millions of posts in the forum. So basically we needed to split those SQL updates into smaller 15-25 second long pieces and to use AJAX as many times as needed.