A false positive is triggered by the linter MultilineScript, so this
explains the switch from =~ to match?.

On top of this, since Ruby 2.4, `String#match?`, `Regexp#match?` and
`Symbol#match?` have been added.  The methods are faster than
`match`/`=~`.  Because the methods avoid creating a `MatchData` object
or saving backref. So, when `MatchData` is not used, use `match?`
instead of `match`/`=~`.

Update selenium-webdriver: 3.13.1 → 3.14.0 (minor)

Update autoprefixer-rails: 9.0.2 → 9.1.0 (minor)

Add Bootstrap version of package#revision

Test Owner search as model spec

Fix issue 5449

Test emoji emails

We didn't have any test covering this method

Before when running PackageController#binary and providing a non
existent architecture the code was raising an exception and resulting in
a 500 in production. Now the code checks if the architecture is valid
before trying to access to its name.

Co-authored-by: David Kang <dkang@suse.com>

split notary functions into BSTUF.pm

Splitting by whitespace and trying to find users is breaking
on 'I love @henne!' - as there is no user 'henne!'

Fix my fix for BsRequestAction_Differ_ForSource

The stub_request was using the project not its name,
which lead to the request appearing in cassettes - which
should have been a warning sign :(

Update capybara: 3.4.2 → 3.5.0 (minor)

They will be used by bs_notar and the publisher

[backend] support setrelease also for kiwi appliances

Fix PublicControllTest to work reliably without VCR

Adopt the expectations of package diff

Request differ: Fix one more test case that is invalid without cassettes

[frontend] Store comments.body as utf8mb4

The cassettes only contained a 404 for a package that the
test didn't create due to missing global_write_through

The bigfile was changed in f59ecf7e,
but the expectations were still following the old cassette

The cassette contains a package that the test never
actually generates - blindly assuming another test
creates it, depending on the order

Allows emojis and fixes #3547

This charset change requires mysql 5.5, which was released 2012
but is still not the default. See more details about the problem
on https://railsmachine.com/articles/2017/05/19/converting-a-rails-database-to-utf8mb4.html

I borrowed the migration from there - but limited it to comments,
the charset is backward compatible, but requires changes to database.yml
in production (to store emojis not to apply the migration)

The old migrations are tested on current database and database
settings. This makes it bound to the current encoding - and that is
utf8mb4. But without Barracuda tables you can't have varchars. So hardcode
the encoding used in 2017 by using execute instead of rails helpers

On create it's easy to add, but renames and other column changes
need to be changed to executes as they otherwise drop the charset

Fix spec file: buildrequire createrepo_c

add tests and cassettes to webui/patchinfo controller

[dist] enforce createrepo_c now

Update autoprefixer-rails: 9.0.0 → 9.0.2 (patch)