test-bot expects the dependencies returned from SoftwareSpec#deps to
satisfy is_a?(TapDependency) if they come from a tap which might need to
be tapped.
Formulas that build python things and which are tested with system
Python will encounter test failures unless system Python is configured
to add Homebrew's site-packages to sys.path. This change makes sure that
configuration is performed in the test environment.
Both lines are needed; the first reads and processes .pth files and the
second makes sure that Homebrew's site-packages is read before the
system extras, so that formulas depending on Homebrew/python/numpy get
the Homebrew/python version and not the old system version.
ClosesHomebrew/homebrew#38466.
This is a more standard way to pass fd in UNIX world.
At the same time, it helps to remove a few hacks and
simplifies the code in the sandbox.
ClosesHomebrew/homebrew#38434.
Signed-off-by: Xu Cheng <xucheng@me.com>
Per requirements.rb:
> XXX If the satisfy block returns a Pathname, then make sure that it
> remains available on the PATH. This makes requirements like
> satisfy { which("executable") }
> work, even under superenv where "executable" wouldn't normally be on the
> PATH.
> This is undocumented magic and it should be removed, but we need to add
> a way to declare path-based requirements that work with superenv first.
Fixeshomebrew/homebrew-python#170.
ClosesHomebrew/homebrew#38448.
Fixes the bottle regex problem seen in
https://github.com/Homebrew/homebrew-versions/pull/678.
I don’t know whether it’s a good regex, or an awful regex, but it works
and passes `brew tests` and a bottled install. Open to improvements if
anyone has them.
ClosesHomebrew/homebrew#38333.
Signed-off-by: Mike McQuaid <mike@mikemcquaid.com>
Comparing PkgVersion and Version objects can produce nonsensical
results. For example, equality is not symmetric:
irb(main):002:0> PkgVersion.new("1.0", 0) == Version.new("1.0")
=> false
irb(main):003:0> Version.new("1.0") == PkgVersion.new("1.0", 0)
=> true
Rather than attempt to deal with subclass-superclass equality, let's use
composition and punt on the problem altogether.
A hyphen is not a valid character in a Ruby symbol, and the bottle
tag should be a valid Ruby symbol for its use in the bottle stanza.
ClosesHomebrew/homebrew#38235.
Signed-off-by: Mike McQuaid <mike@mikemcquaid.com>
Change [a-z_]+(32)? to [-a-z0-9_]+
The Linuxbrew bottle tag is x86_64-linux.
ClosesHomebrew/homebrew#32687.
Signed-off-by: Mike McQuaid <mike@mikemcquaid.com>
test.pdf wasn't a valid PDF file:
>$ mutool info test.pdf
>test.pdf:
>error: cannot recognize xref format
>error: cannot read xref (ofs=406)
>error: cannot read xref at offset 406
>warning: trying to repair broken xref
Process it with `mutool clean` to fix errors, collapse newlines to reduce size
ClosesHomebrew/homebrew#38194.
Signed-off-by: Xu Cheng <xucheng@me.com>
Packages supporting CMake may install configuration files for use
with CMake's find_package command. A recommended location for these
is `<prefix>/lib/cmake` which is not unique across packages. This
may cause issues for Formula using this location when their Keg
is linked.
As with pkg-config, explicitly create the `lib/cmake` folder when
linking a Keg that has installed folders/files to this location.
Also add testcase.
ClosesHomebrew/homebrew#38005.
Signed-off-by: Xu Cheng <xucheng@me.com>
If the formula name is in full-qualified name. Let's silently
ignore it as we don't care about things used in taps that aren't
currently tapped.
ClosesHomebrew/homebrew#38089.
Signed-off-by: Xu Cheng <xucheng@me.com>
This is a follow-up of Homebrew/homebrew#37734. I think we should treat `install_metafiles` as
a general issue rather than a formula-specific one.
ClosesHomebrew/homebrew#37968.
Signed-off-by: Xu Cheng <xucheng@me.com>
