Procs cannot be marshalled, but formula objects may be as part of the
BuildError exception. If the formula object has a reference to a Proc,
this will fail.
Work around it by not storing the test Procs during installations.
Tests can now be specified as a block in the DSL. A temporary test
directory is set up automatically when calling Formula#test. The
semantics of the test remain the same: the block can either raise an
exception or return false to signal failure.
Signed-off-by: Jack Nagel <jacknagel@gmail.com>
This code makes assumptions about the existence of prefix which are
valid in the context of the installer, but not necessarily in the
context of `brew info`, thus `brew info` on an outdated formula errors
out.
This reverts commit e5b53dd64b769b67805d1054d906f7083939d905.
Formula.factory naively assumed that any already-loaded constant
fed into it was a formula, with confusing results when trying to
`brew install` a name from the ruby stdlib.
FixesHomebrew/homebrew#16284.
The new formula debugger users continuations to jump back up the stack,
and thus anything that has the potential to be executed twice (i.e.
ensure blocks) needs to to be essentially idempotent.
Signed-off-by: Jack Nagel <jacknagel@gmail.com>
A new feature for easing the pain of working with complex formulas, or
formulas for large packages. When running brew in debug mode (-d), if an
exception propagates outside the formula's install method, you now get a menu
which lets you return to the point where the exception was raised and perfom
several useful actions, such as:
- printing a backtrace
- entering IRB to examine the context and test ruby code
- entering the debugger (if ruby-debug is available)
- entering a shell
- ignoring the exception or proceeding with the raise as normal
Signed-off-by: Max Howell <mxcl@me.com>
* Fixed conflict in build.rb.
* Removed old debug handling in Formula.brew.
ClosesHomebrew/homebrew#10435.
ComparableSet only allows a single object of a given class,
choosing the object with the greatest value. This was mainly created
for Requirements, so that, e.g., two X11Dependencies of differing
strictness don't both end up in the same requirement set.
FixesHomebrew/homebrew#15240.
This started happening as a result of bbc167b, which avoids reloading
previously defined formulae. The exception alters its message if the
name parameter appears to be a tapped formula, but this is lost if we
only pass in the "stem" of the pathname.
Long-term this probably needs some restructuring, perhaps with a
separate exception for tapped formula, but for now this should re-enable
the more useful error message.
FixesHomebrew/homebrew#15374.
Signed-off-by: Jack Nagel <jacknagel@gmail.com>
Setting $VERBOSE in the environment makes the haskell-platform build
fall down, which is surely one of the odder bugs I've run into.
See Homebrew/homebrew#15164.
All logs are now stored from each command executed in Formula.install.
Error output is truncated to five lines in an attempt to not overwhelm the user and to encourage users to read the error output and report the bug properly. Maybe we can get that figure up from 70% to 90%.
Output JSON with the --json=v1 option. Output is in an array, and
supports one or more formulae (or all, with the --all option).
Why 'v1'? The format is unstable, presumably we'll deprecate it
someday. It should be solid by Homebrew 1.0.
ClosesHomebrew/homebrew#13299.
This patch removes most of the settings for CC, CXX etc. because we are trying to be minimal. Then we force the compiler to Homebrew's choice underneath in superenv. We however leave LD because we prefer that build-systems use the c-compiler for linking, it generally works better (copiously tested), however when the build-system explicitly calls ld, we respect that. This gets around the ocaml bug in question, since somehow clang was crashing during link, but the ld tool itself (which is kind of clang, kind of llvm-gcc) is okay with this.
Also moved the setting of O (so that cc-args are refurbished) into a make wrapper. Not sure if this matter much, but seems more consistent.
This was mostly used to prevent stripping. Now we don't strip. Force formula to remove this setting or be more specific about folders they don't want pruned.
1. A minimal build environment, we don't set CFLAGS, CPPFLAGS, LDFLAGS, etc. the rationale being, the less that is set, the less variables we are introducing that can break builds.
2. A set of scripts that replace cc, ld, etc. and inject the -I, -L, etc. flags we need into the args passed to the build-tools.
Because we now have complete control over compiler instantiations we do a variety of clean-up tasks, like removing bad flags, enforcing universal builds and ensuring makefiles don't try to change the order of library and include paths from ones that work to ones that don't.
The previous ENV-system is still available when --env=std is specified.
superenv applies to Xcode >= 4.3 only currently.
Similar to the LinkedKegs record, we write a symlink for installed kegs to PREFIX/opt.
Unlike the linked-keg record, unlinking doesn't remove the link, only uninstalling, and keg-only formula have a record too.
The reason for this addition is so that formula that depend on keg-only formula can build against the opt directory and not the cellar keg. Thus surviving upgrades.
To enforce this fix_install_names and built were adapted to use the opt path.
Standard kegs also create an opt symlink so that caveats can now refer to the opt directory and thus provide steps that survive upgrades too.
Thus the choice of /opt. It is short, neat and the right choice: POSIX dictates that opt is for stand-alone prefixes of software.
Specifying dependencies with a URL works, even if by accident, but
factory is called repeatedly on this URL and this results in multiple
downloads of the same file.
Fix this by checking const_defined? here too, and DRY up the code a bit.
FixesHomebrew/homebrew#14285.
Signed-off-by: Jack Nagel <jacknagel@gmail.com>
Due to the new const_defined? checks, passing certain names (such as
those that start with a digit) that are illegal constant names will
raise NameError. Catch this and error out gracefully, as we would have
previously.
FixesHomebrew/homebrew#14342.
Signed-off-by: Jack Nagel <jacknagel@gmail.com>
build.rb calls Formula.factory to get a usable Formula object to pass to
its install method. However, because the formula file is the actual
executing script, its class is already defined, and loading it again
causes the class to be re-evaluated, which, unfortunately, is not
idempotent.
This bug has existed for a very long time, and its side effects include
duplicate entries the deps array and mirrors array, among others.
Fortunately, the fix is very simple.
Signed-off-by: Jack Nagel <jacknagel@gmail.com>
Usually, the "foo-version already installed" error is printed by
FormulaInstaller. However, if an up-to-date formula that has outdated
deps is passed on the command line, we proceed to upgrade the deps and
then print a message saying that the formulae given on the command line
is already installed.
Catch this earlier, when the outdated list is being populated, print an
appropriate message, and skip the up-to-date formula.
Signed-off-by: Jack Nagel <jacknagel@gmail.com>
