This got updated recently in 42a42c96acc5193dbb6e321ff2aaecceb00e48a4
to split out the livecheck audits and as a result of that the type
signature for the #validate_url_for_https_availablity method got updated.
This did not account for the possibility that the livecheck url was nil.
I've added a nil check here since it makes no sense to try and validate
a nil url and we have nil checks for the other two http availability
audits already. Plus, the livecheck blocks are audited thoroughly already
for syntax.
We'd like to reduce the size of the API JSON and to do that we are
going to remove unused and/or blank elements from the cask/formula
definition. This will reduce the amount of data that has to go
over the wire and make it easier to load this data into memory.
We were only loading casks from files with this command which,
of course, didn't work for casks that can only be loaded from
the API (when the core cask tap is not tapped). This changes
things to align more with what we do for formulae.
At the moment, Microsoft Office and related casks fail audit due to
failure to fetch a product homepage from the main Microsoft site
(https://www.microsoft.com/) served by Akamai CDN.
The failure is severe and weird: no status code is received as the
connection is reset.
The analysis revealed the issue is linked to `User-Agent` and
`Accept-Language` headers parsing. Homebrew uses involved user agent
strings and has language hardcoded as `en`.
The simplest workaround is to use a simple user agent string when checking
homepage availability if the cask is in an audit exceptions list.
Merging this would fix Microsoft Office updates (including
https://github.com/Homebrew/homebrew-cask/pull/162671)
- As discussed in
https://github.com/Homebrew/brew/pull/16337#issuecomment-1855668516,
then we won't have the chicken/egg problem of fixing cops for syntax
that's unsupported on Ruby 2.6 _before_ we set the TargetRubyVersion
to 3.1, but if we set it too early then all the taps will fail.
If an installed cask is invalid on attempting an untap: it will
prevent untapping that cask.
Fix this in two ways: one more specific to `untap` and one more
generally to other commands too:
- specific: only read the actual formulae/casks from the tap we're
untapping instead of all of those that are installed
- general: rescue more exceptions in `Cask::Caskroom.casks` (like we
already do for `Formula.installed`
This was not returning the full name correctly for e.g. anything in
Homebrew/homebrew-fonts.
While we're here, fix up a few other places where `tap.core_cask_tap?`
can be used more appropriately.
This change doesn't affect the behavior of the `#audit_min_os`
method and simply reorders the array members to place `plist_min_os`
before `sparkle_min_os` for the sake of consistency (using the same
order as the preceding lines).
The `#livecheck_min_os` cask audit method manually replicates some of
the `Sparkle` strategy's behavior but in an incomplete way that has
lead to inappropriate audit failures at times. This reimplements it
to use `Livecheck` methods, so it will align with the `Sparkle`
strategy's behavior.
The `#livecheck_min_os` cask audit method should be skipped when a
`Sparkle` `livecheck` block contains a `strategy` block that uses
the `items` argument (instead of `item`). These `strategy` blocks
contain arbitrary logic that ignores/overrides the strategy's sorting,
so we can't identify which item would be first/newest.
