Basically, const_missing had a loop to try parent namespaces
if the constant lookup failed, but at the same time delegated
to load_missing_constant which in turn also walks up parent
namespaces calling const_missing by hand. In the case of missing
constants this results in repeated work in some funky nested way.
When a block is passed into the method, it will be invoked for each
duplicated key, with the key in question and the two values as
arguments. The value for the duplicated key in the receiver will
be set to the return value of the block.
This behaviour matches Ruby's long-standing implementation of
Hash#update and is intended to provide a more consistent interface.
HashWithIndifferentAccess#merge is also affected by the change, as it
uses #update internally.
We need to anchor to remove the extension. In addition to
be the correct way to do that, files in ~/.rbenv get that
.rb removed, so it is a real source of bugs, as reported in
b33700f558 (commitcomment-1781840)
Nowadays circular autoloads do not work, but the user gets a NameError
that says some constant is undefined. That's puzzling, because he is
normally trying to autoload a constant he knows can be autoloaded.
With this check we can give a better error message.
loaded stores file names without the .rb extension, but search_for_file
returns file names with the extension.
The solution is hackish, but this file needs a revamp.
We simplify two things here: First since * is greedy it is enough to go
look for the rightmost ::, no need to ask the regexp engine to match the
rest of the string since we are not validating anything, only capturing.
The second simplification comes from using a look-ahead assertion, that
allows us to have the capture in $&, thus removing the need of a group.
This reverts commit b0ab8dc0b2b0f580ffe5ac9ff57fd13152e18577
because it was removing the contents of the message when we
did not have any tag. A test case is also committed.
The revised test assumed that the default permissions of a file
matched the umask of the process, but in the general case that
depends also on the file system. This test was failing in the
/vagrant shared folder of Rails development boxes.
The new option allows any Ruby namespace to be registered and set
up for eager load. We are effectively exposing the structure existing
in Rails since v3.0 for all developers in order to make their applications
thread-safe and CoW friendly.
Previously, ActiveSupport::Autoload was global and reserved
for usage inside Rails. This pull request makes it local,
fixes its test (they were not being run because its file
was named wrongly) and make it part of Rails public API.
Ruby does not pass the nesting to const_missing (unfortunately).
That second argument was there in case that changed, Yehuda
sent a patch to MRI
http://bugs.ruby-lang.org/issues/2740
but there is not much movement there and Matz told me in
Amsterdam there was no immediate plan to pass the nesting.
So let's go back to implement what happens now, and if
in the future we get the nesting then we will adapt this.
Double-checked this with Mr Katz.
In cases where a notification subscriber includes methods to support
both Evented and Timed events, Evented should take priority over Timed.
This allows subscribers to be backwards compatible (older Rails only
allows Timed events) while defaulting to newer behavior.
Always merge I18n format values, namespaced or not, over the default
ones, to ensure I18n format defaults will have precedence over our
namespaced values.
Precedence should happen like this:
default :format
default :namespace :format
i18n :format
i18n :namespace :format
Because we cannot allow our namespaced default to override a I18n
:format config - ie precision in I18n :format should always have higher
precedence than our default precision for a particular :namespace.
Also simplify default format options logic.
Action Pack already comes with a default locale fine for :en, that is
always loaded. We can just fallback to this locale for defaults, if
values for the current locale cannot be found.
Closes#4420, #2802, #2890.
Reason: ActiveSupport::JSON::Variable is not used anymore internally. It
was deprecated in 3-2-stable but we reverted all the deprecation for
point releases.
See #6536 and #6546.
Conflicts:
activesupport/lib/active_support/json/variable.rb
Selecting which key extensions to include in active_support/rails
made apparent the systematic usage of Object#in? in the code base.
After some discussion in
5ea6b0df9a
we decided to remove it and use plain Ruby, which seems enough
for this particular idiom.
In this commit the refactor has been made case by case. Sometimes
include? is the natural alternative, others a simple || is the
way you actually spell the condition in your head, others a case
statement seems more appropriate. I have chosen the one I liked
the most in each case.
* Added the `DateAndTime::Calculations` module that is included in Time
and Date. It houses common calculations to reduce duplicated code.
* Simplified and cleaned-up the calculation code.
* Removed duplication in tests by adding a behavior module for shared
tests. I also added some missing tests.
