aquarium 0.5.1 → 0.7.1

checksums.yaml

@@ -0,0 +1,7 @@
+---
+SHA256:
+  metadata.gz: 607c8e2ff478a03b28cb82786e11db87f2809aeedb30749ca57c93e3da75a8e3
+  data.tar.gz: 027f4f0635c170239546fde6547f8c84a684020046cd940f3466ab9f8bbb45a4
+SHA512:
+  metadata.gz: de21011f1c22b93eac2de02630836c40759511a80d2d02b6ee626e98a17b41bcaf50d1bb65393da9123b1638e7b503c37075764e30fbb380dd25e216134ac6e3
+  data.tar.gz: 30157999d9b464b2bdcb8808a14ab7f1ed90b89a8959f7cc9fe5ed7cb52f4355c6a4770f6210622d47b3e2ec85fe6e8f80ab917eb74a26d2cedd5d3448fc911f

data/CHANGES

@@ -1,12 +1,37 @@
+== Version 0.7.1
+
+Adds support for Ruby 2.6. It was tested with 2.6.3p62 (2019-04-16 revision 67580). Basic JRuby 9.2.7.0 (2.5.3) 2019-04-09 8a269e3 support is provided, but using JRuby currently does *not* pass all the custom Java specs in the `jruby/spec` directory, which specifically test working with Java classes. However, JRuby does pass all the Ruby specs in the `spec` directory. This is a long-standing issue with JRuby support in Aquarium. Patches are welcome!
+
+== Version 0.6.0
+
+V0.6.0 fixes thread-unsafe code in Aspect::alias_original_method_text, which
+I've known about since the beginning of the project. At the beginning of Aquarium,
+circa 2006, the overhead of making copies of JoinPoints in this method was
+approximately 30%! So, I sacrificed thread-safety for performance. Now, thread
+safety is increasingly important in modern apps, and fortunately, the overhead
+on newer machines and newer versions of Ruby is greatly reduced for making these
+copies. Hence, this problem is fixed.
+
+The rspec suite does not test thread safety in any rigorous way, so caution is
+advised. Please report any problems or better, submit pull requests of better
+threading tests and possible bug fixes. As always, thanks in advance.
+
+I also fixed deprecation warnings from the latest version of RSpec. There were a
+lot of them!
+
+Now supports Ruby 1.9.X, Ruby 2.0, and JRuby 1.7.4 (please report version-related bugs!)
+Drops official support for Ruby 1.8.X.
+
+
 == Version 0.5.1
 
 V0.5.1 adds support for Ruby 1.8.7, Ruby 1.9.3, JRuby 1.6.7,
 RSpec 2.9, and Webgen (0.5.14 - for the build process).
 
-WARNING: Earlier versions of Ruby 1.8, 1.9 and JRuby 1.X are not 
+WARNING: Earlier versions of Ruby 1.8, 1.9 and JRuby 1.X are not
 supported. JRuby support is limited to "pure" Ruby code; the main
-RSpec specifications pass under JRuby, but not the JRuby-specific 
-specifications in jruby/spec. Restoring full JRuby support is planned, 
+RSpec specifications pass under JRuby, but not the JRuby-specific
+specifications in jruby/spec. Restoring full JRuby support is planned,
 but the timeframe is TBD.
 
 It also has some minor API changes summarized below.
@@ -16,28 +41,28 @@ Enhancements:
 27235	Support RSpec 1.2.8 (actually 2.9) and and JRuby 1.3.1 (actually 1.6.7)
 27236	Upgrade website generation to webgen 0.5.X (X=14)
 
-The actual versions of these tools supported are more recent versions 
+The actual versions of these tools supported are more recent versions
 shown in parentheses.
 
-There are no API changes in this release, with two exceptions. 
+There are no API changes in this release, with two exceptions.
 
-First, Most of the code changes are minor fixes to RSpec files. The one 
-exception is an apparent change that occurred in JRuby 1.3.X with respect 
-to the access restrictions on Java methods. It now appears that protected 
-methods are public, that is they are in the method list returned by 
-:public_methods, but not :protected_methods. Also, private methods are 
-not returned by any of the :*_methods. 
+First, Most of the code changes are minor fixes to RSpec files. The one
+exception is an apparent change that occurred in JRuby 1.3.X with respect
+to the access restrictions on Java methods. It now appears that protected
+methods are public, that is they are in the method list returned by
+:public_methods, but not :protected_methods. Also, private methods are
+not returned by any of the :*_methods.
 
-Second, there was a deprecated option to the MethodFinder class, 
+Second, there was a deprecated option to the MethodFinder class,
 :options, that has been removed. Use :method_options instead.
 
-I used the upgrade to Webgen 0.5.X as an excuse to refresh the web site's 
+I used the upgrade to Webgen 0.5.X as an excuse to refresh the web site's
 appearance. The content is mostly the same as before.
 
 
 == Version 0.4.3
 
-V0.4.3 adds a few bug fixes and enhancements, and several internal 
+V0.4.3 adds a few bug fixes and enhancements, and several internal
 refactorings to improve performance, etc.
 
 Bug fixes:
@@ -45,114 +70,114 @@ Bug fixes:
 
 Enhancements:
 14165	Support Ruby 1.8.2, 1.8.5, 1.8.6, 1.9.X and 2.0 releases and snapshots
-21283	Allow "_of" suffix at end of "*_and_descendents", "*_and_ancestors", 
+21283	Allow "_of" suffix at end of "*_and_descendents", "*_and_ancestors",
   and "*_nested_types"
-21924	Should be able to turn off contract-testing aspects globally in the 
+21924	Should be able to turn off contract-testing aspects globally in the
   Design by Contract extra module
-21993	By default, don't advise "system" methods that being with "_", but 
+21993	By default, don't advise "system" methods that being with "_", but
   provide an option to do so
 
 Details:
-14165: I closed this one because it's too broad. Aquarium currently runs with 
+14165: I closed this one because it's too broad. Aquarium currently runs with
   1.8.6. I will open a new enhancement specifically for 1.9.
-21283: Sometimes appending "_of" makes the specification read more smoothly, 
+21283: Sometimes appending "_of" makes the specification read more smoothly,
   so it's now an option.
-21924: To class-level methods in DesignByContract, enable_all and disable_all, 
+21924: To class-level methods in DesignByContract, enable_all and disable_all,
   globally turn contracts on or off. See their documentation for details.
-21993: Now, by default, MethodFinder won't match any method that starts with 
-  two underscores ("__"), so you don't have to :exclude_ancestors as much just 
-  to avoid matching methods like  "__id__" and "__send__". There is also a new 
-  :method_options called :include_system_methods that will not suppress 
-  matching these methods. So, I'm calling these special methods "system 
-  methods". I don't yet provide a way to configure this list (RSpec methods 
-  would be good additions...). The list is currently defined as an array of 
-  regex's, MethodFinder::IGNORED_SYSTEM_METHODS. NOTE: This is effectively a 
+21993: Now, by default, MethodFinder won't match any method that starts with
+  two underscores ("__"), so you don't have to :exclude_ancestors as much just
+  to avoid matching methods like  "__id__" and "__send__". There is also a new
+  :method_options called :include_system_methods that will not suppress
+  matching these methods. So, I'm calling these special methods "system
+  methods". I don't yet provide a way to configure this list (RSpec methods
+  would be good additions...). The list is currently defined as an array of
+  regex's, MethodFinder::IGNORED_SYSTEM_METHODS. NOTE: This is effectively a
   behavior change, although it's not likely to affect anyone.
 
 == Version 0.4.2
 
-V0.4.2 adds a few bug fixes and enhancements, greatly improved RDoc output, 
+V0.4.2 adds a few bug fixes and enhancements, greatly improved RDoc output,
 and internal refactorings to improve performance.
 
 Bug fixes:
 15202	Intermittent confusion between classes and objects when invoking advice
-19262	Just putting join_point argument in advice block causes mysterious 
+19262	Just putting join_point argument in advice block causes mysterious
   method missing error.
 19321	Advice overhead is too high (ongoing improvements)
 
 Enhancements:
-13403	Support recursion into nested types without requiring "tricky" regular 
+13403	Support recursion into nested types without requiring "tricky" regular
   expressions
 13406	"Sugar" for adding methods and attributes to types
 18537	Provide a search for all pointcut constants
 19666	Improve the RDoc output
-19119	Provide an "after_raising" type of advice that lets you handle the 
+19119	Provide an "after_raising" type of advice that lets you handle the
   exception.
 
-#15202: I never figured out the cause of this problem, but it hasn't been seen 
-  since late last 
-year, so I suspect it disappeared as a side effect of on-going refactoring and 
+#15202: I never figured out the cause of this problem, but it hasn't been seen
+  since late last
+year, so I suspect it disappeared as a side effect of on-going refactoring and
   enhancements.
 
-#19262: If you just specified "|jp|" for an advice block, you would sometimes 
-  get a method missing error. I never figured out exactly why, but it was 
-  somehow related to passing the usual three arguments internally, where the 
-  last two would be ignored in this case. Now, the code checks the arity and 
-  only passes the join point in this case.  
+#19262: If you just specified "|jp|" for an advice block, you would sometimes
+  get a method missing error. I never figured out exactly why, but it was
+  somehow related to passing the usual three arguments internally, where the
+  last two would be ignored in this case. Now, the code checks the arity and
+  only passes the join point in this case.
 
-#19321: I removed some of the wasted object creation and initialization in 
-  advice invocation, 
-improving the overhead by about 40%. However, it is still at least 10x higher 
-  than simple method aliasing, so I want to make more improvements. (I did not 
+#19321: I removed some of the wasted object creation and initialization in
+  advice invocation,
+improving the overhead by about 40%. However, it is still at least 10x higher
+  than simple method aliasing, so I want to make more improvements. (I did not
   close this item.)
 
-#13403: I added new options :types_and_nested_types and :types_and_nested that 
-  are analogous to the similar "ancestors" and "descendents" options. The 
-  nested option will return the specified types and any types they "enclose". 
+#13403: I added new options :types_and_nested_types and :types_and_nested that
+  are analogous to the similar "ancestors" and "descendents" options. The
+  nested option will return the specified types and any types they "enclose".
   There are also corresponding "exclude" options.
 
-#13406: I've decided not to do this, as it really isn't the "simplest thing 
+#13406: I've decided not to do this, as it really isn't the "simplest thing
   that could possibly work." It's easy enough for the user to define a module
-  of new behavior and just use "obj.extend(module)". However, when the user 
+  of new behavior and just use "obj.extend(module)". However, when the user
   needs to do this over a set of types, Aquarium's TypeFinder can be helpful,
   so I added an example to the Examples code and page.
 
-#18537: I've provided an example of the design approach where you define 
-  pointcuts in a class, as a kind of "aspect interface" and write aspects that 
-  specify those pointcuts. The problem has been that you had to name every 
-  single such pointcut explicitly. There was no "finder" option, as for types, 
-  methods, etc. Now there is a pointcut finder option with a new option 
-  ":named_pointcuts" for Aspect.new to specify a search for pointcuts in a set 
-  of types matching a name specification. Either constants or class variables 
+#18537: I've provided an example of the design approach where you define
+  pointcuts in a class, as a kind of "aspect interface" and write aspects that
+  specify those pointcuts. The problem has been that you had to name every
+  single such pointcut explicitly. There was no "finder" option, as for types,
+  methods, etc. Now there is a pointcut finder option with a new option
+  ":named_pointcuts" for Aspect.new to specify a search for pointcuts in a set
+  of types matching a name specification. Either constants or class variables
   will be matched (or both).
 
-#19666: The rdoc for the key classes was cleaned up so it "renders" better. 
+#19666: The rdoc for the key classes was cleaned up so it "renders" better.
   Feedback is welcome.
 
-#19119: I finished the previously-incomplete support for allowing advice to 
-  change the raised exception, in both after and after_raising advice. A 
-  common scenario is to wrap the thrown exception in another. For example, a 
+#19119: I finished the previously-incomplete support for allowing advice to
+  change the raised exception, in both after and after_raising advice. A
+  common scenario is to wrap the thrown exception in another. For example, a
   low-level, service-specific exception (like a database error) in a
-  higher-level, more generic application exception. 
+  higher-level, more generic application exception.
 
-You still can't rescue the exception completely in :after_raising and :after 
-advice; the value for the exception in 
+You still can't rescue the exception completely in :after_raising and :after
+advice; the value for the exception in
   joinpoint.context.raised_exception
-when the advice returns will be raised by Aquarium. I think that permitting 
-:after_raising or :after advice to "eat" the exception could cause subtle 
-issues with scope and variable binding. It would also probably violate the 
-"principle of least surprise"; the advice code that rescues the exception would 
+when the advice returns will be raised by Aquarium. I think that permitting
+:after_raising or :after advice to "eat" the exception could cause subtle
+issues with scope and variable binding. It would also probably violate the
+"principle of least surprise"; the advice code that rescues the exception would
 not be as "obvious" to the reader as the familiar idiom of rescue clauses
-that we are accustomed to using. Therefore, if you want to recover completely 
+that we are accustomed to using. Therefore, if you want to recover completely
 from an exception, use rescue clauses in around advice.
 
 == Version 0.4.1
 
-V0.4.1 adds a few bug fixes, a few more user examples, internal refactoring and 
+V0.4.1 adds a few bug fixes, a few more user examples, internal refactoring and
 some performance improvements.
 
 Bug fixes:
-19116	When an exception is thrown during advice execution, the error message 
+19116	When an exception is thrown during advice execution, the error message
   always reports the advice type is :before!
 19261	after_raising DSL method provides no way to specify exceptions
 
@@ -161,19 +186,19 @@ Enhancements:
 19320	Move the Aquarium::...::AspectDSL file to Aquarium::DSL for convenience
 19399	Improve the Design by Contract example
 
-I added a new :exceptions argument (synonym :exception) that takes a single 
-exception or list thereof. You can only use this argument with :after_raising. 
-If you specify exceptions with the latter and use the :exceptions argument, 
+I added a new :exceptions argument (synonym :exception) that takes a single
+exception or list thereof. You can only use this argument with :after_raising.
+If you specify exceptions with the latter and use the :exceptions argument,
 the values will be merged.
 
-I thought it was ugly to have to type "include 
-Aquarium::Aspects::DSL::AspectDSL", so I moved the code so now it's "include 
-Aquarium::DSL". However, for backwards compatibility, the old module still 
+I thought it was ugly to have to type "include
+Aquarium::Aspects::DSL::AspectDSL", so I moved the code so now it's "include
+Aquarium::DSL". However, for backwards compatibility, the old module still
 works.
 
 == Version 0.4.0
 
-V0.4.0 adds specs to characterize and test advising Java classes when running 
+V0.4.0 adds specs to characterize and test advising Java classes when running
 on JRuby and adds several API enhancements.
 
 Bug fixes:
@@ -185,50 +210,50 @@ Enhancements:
 17834	Allow :class and :module (and variants) wherever :type is allowed
 17881	Add specs that exercise advising Java types and objects using JRuby
 
-#17844 occurred because of the way JRuby encodes Java packages into modules. 
+#17844 occurred because of the way JRuby encodes Java packages into modules.
   Aquarium now properly handles JRuby types.
 
 #17883 reflected a JRuby bug, so a workaround was required.
- 
-#17834 allows the user to substitute the words "class", "classes", "module" or 
-"modules" anywhere the words "type" and "types" are used in the API, since some 
+
+#17834 allows the user to substitute the words "class", "classes", "module" or
+"modules" anywhere the words "type" and "types" are used in the API, since some
 users might naturally want to write aspects like this:
 
  	around :calls_to => :my_method, :in_class => MyClass do ...; end
 
-However, there is no enforcement to ensure that "class" is only used for 
-classes and "module" is only used for modules, etc. Note: it's possible we'll 
-enforce this in some future release, as a way of saying things like "only advise 
+However, there is no enforcement to ensure that "class" is only used for
+classes and "module" is only used for modules, etc. Note: it's possible we'll
+enforce this in some future release, as a way of saying things like "only advise
 classes that match ...", etc. Caveat Emptor!
 
-For #18090, a bug prevented JoinPoint#invoke_original_join_point (which allows 
-you to bypass all advices at the join point) from working except for :around 
+For #18090, a bug prevented JoinPoint#invoke_original_join_point (which allows
+you to bypass all advices at the join point) from working except for :around
 advice (and yes, the specs didn't cover this adequately - gasp!). Now fixed.
 
-For #17881, I created a separate set of specs for JRuby, so it's easier to run 
-the "regular" Aquarium specs using MRI and the JRuby-specific specs separately 
-with JRuby. The new "jruby" directory contains a Rakefile, another set of specs, 
-and Java example code for the specs to use. The default Rakefile task re-runs 
-the main Aquarium spec suite using JRuby, to confirm that the suite passes 
-successfully, then it runs a set of different specs that load sample Java 
+For #17881, I created a separate set of specs for JRuby, so it's easier to run
+the "regular" Aquarium specs using MRI and the JRuby-specific specs separately
+with JRuby. The new "jruby" directory contains a Rakefile, another set of specs,
+and Java example code for the specs to use. The default Rakefile task re-runs
+the main Aquarium spec suite using JRuby, to confirm that the suite passes
+successfully, then it runs a set of different specs that load sample Java
 interfaces and classes into JRuby and then advises them with Aquarium.
 
-I found a few JRuby bugs and other behavior differences between MRI during this 
-release. I was able to handle them with modifications to the Aquarium code. If 
-you "grep" the Aquarium "lib" and "spec" directories for the word "jruby" 
-(ignoring case), you'll find notes about these issues and the workarounds I 
+I found a few JRuby bugs and other behavior differences between MRI during this
+release. I was able to handle them with modifications to the Aquarium code. If
+you "grep" the Aquarium "lib" and "spec" directories for the word "jruby"
+(ignoring case), you'll find notes about these issues and the workarounds I
 implemented.
 
-The separate JRuby spec suite shows what you can and cannot do with Java types. 
-As a side benefit, it also demonstrates how Java types, objects, and methods 
-are mapped to Ruby. There are some important limitations, however. See the 
-jruby.html page on the website or the README for more details. 
+The separate JRuby spec suite shows what you can and cannot do with Java types.
+As a side benefit, it also demonstrates how Java types, objects, and methods
+are mapped to Ruby. There are some important limitations, however. See the
+jruby.html page on the website or the README for more details.
 
 Note: JRuby 1.1RC2 was used.
- 
+
 == Version 0.3.1
 
-V0.3.1 adds numerous performance improvements, especially in the RSpec suite, 
+V0.3.1 adds numerous performance improvements, especially in the RSpec suite,
 and some minor API additions.
 
 Bug fixes:
@@ -236,60 +261,60 @@ N.A.
 
 Enhancements:
 14447	Unify internal handling of reporting errors to the user
-17514	Provide an Aquarium library-wide logger with configuration parameters and 
+17514	Provide an Aquarium library-wide logger with configuration parameters and
   instance overrides
 17515	Add an optional warning if an aspect doesn't match any join points
-17516	Remove unnecessary examples that use :types_and_descendents to shorten 
+17516	Remove unnecessary examples that use :types_and_descendents to shorten
   time to run the RSpec suite
-17565	JoinPoint.new should convert a type name, symbol, or regex to the type 
+17565	JoinPoint.new should convert a type name, symbol, or regex to the type
   and only allow one type
 
-These first two enhancements are related. There is a now an 
-Aquarium::Utils::DefaultLogger module with static accessors for getting and 
-setting the "system-wide" logger. 
+These first two enhancements are related. There is a now an
+Aquarium::Utils::DefaultLogger module with static accessors for getting and
+setting the "system-wide" logger.
 
-When instance-level overrides are necessary, the Aquarium::Utils::OptionsUtils 
-provides "universal" options (but currently used only by Aspect and Pointcut) 
-for specifying a logger (with the new :logger parameter), or alternatively, 
-specifying just the output stream (:logger_stream) and/or the severity 
-(:severity, one of the standard library's Logger::Severity-defined constants). 
-If either of the latter two options is specified, a separate logger instance is 
+When instance-level overrides are necessary, the Aquarium::Utils::OptionsUtils
+provides "universal" options (but currently used only by Aspect and Pointcut)
+for specifying a logger (with the new :logger parameter), or alternatively,
+specifying just the output stream (:logger_stream) and/or the severity
+(:severity, one of the standard library's Logger::Severity-defined constants).
+If either of the latter two options is specified, a separate logger instance is
 created, rather than changing parameters for the global logger.
 
-OptionsUtils also supports a :noop parameter, which classes interpret to mean 
+OptionsUtils also supports a :noop parameter, which classes interpret to mean
 do none (or perhaps only some) of the processing. Useful for debugging.
 
-#17515 adds a helpful warning to the system (or aspect-instance's) logger if 
-  an aspect matches no join points. This warning will be suppressed if (i) the 
-  severity level for the logger is above WARN or (ii) the aspect was created 
+#17515 adds a helpful warning to the system (or aspect-instance's) logger if
+  an aspect matches no join points. This warning will be suppressed if (i) the
+  severity level for the logger is above WARN or (ii) the aspect was created
   with the option :ignore_no_matching_join_points => true.
 
-#17516 fixes halved the long execution times for the whole RSpec suite by 
-refactoring some examples that used type finders with the :types_and_descendents 
-option unnecessarily. It is a very intensive computation! Note that I stubbed 
-out these calls using an aspect with around advise, a useful "development-time" 
+#17516 fixes halved the long execution times for the whole RSpec suite by
+refactoring some examples that used type finders with the :types_and_descendents
+option unnecessarily. It is a very intensive computation! Note that I stubbed
+out these calls using an aspect with around advise, a useful "development-time"
 aspect. See Aquarium::TypeUtilsStub (in spec_example_types.rb) and how it's used
-in pointcut_spec.rb. Using this technique and other optimizations, the time to 
+in pointcut_spec.rb. Using this technique and other optimizations, the time to
 run the suite was reduced from ~5 minutes to about 1 minute.
- 
-#17565 fixes a "hole" in JoinPoint, where it doesn't confirm that a specified 
-type string, symbol or regex matches a class that exists and only one class. Now 
+
+#17565 fixes a "hole" in JoinPoint, where it doesn't confirm that a specified
+type string, symbol or regex matches a class that exists and only one class. Now
 it does and it stores the type, rather than the original "specification" for it.
 
 == Version 0.3.0
 
-V0.3.0 adds numerous improvements to the DSL, making aspect specification more 
+V0.3.0 adds numerous improvements to the DSL, making aspect specification more
 intuitive and English-like. For example, where you previously wrote, e.g.,
 
 	around :methods => :all, :types => [Foo, Bar], :advice => advice_proc
 	after  :attribute => name, :attribute_options => [:readers], :objects => [foo, bar] ...
-	
+
 Now you can write the same aspects as follows:
 
 	around :calls_to => :all_methods, :within_types => [Foo, Bar], :use_advice => advice_proc
 	after  :reading => name, :on_objects => [foo, bar] ...
 
-Other improvements include performance and robustness enhancements and 
+Other improvements include performance and robustness enhancements and
 miscellaneous internal refactoring and DRY improvements.
 
 Bug fixes:
@@ -300,23 +325,23 @@ Enhancements:
 
 For #17154, the following changes were made:
 Added :all_methods as a synonym for the :all special value.
-Added :reading as a synonym for :attributes => ..., :attribute_options => 
+Added :reading as a synonym for :attributes => ..., :attribute_options =>
   [:readers]
-Added :writing and :changing as synonyms for :attributes => ...,   
+Added :writing and :changing as synonyms for :attributes => ...,
   :attribute_options => [:writers]
 Added :accessing as a synonym for :attributes => ...
-Added :calls_to, :calling, :invoking, :sending_message_to  as synonyms for  
+Added :calls_to, :calling, :invoking, :sending_message_to  as synonyms for
   :methods.
-Added :on_types, :in_types, :within_types and :for_types as synonyms for :types.   
-  The same set of prefixes is supported for :type, :objects, :object, and the 
+Added :on_types, :in_types, :within_types and :for_types as synonyms for :types.
+  The same set of prefixes is supported for :type, :objects, :object, and the
   various :exclude_*, :*_and_ancestors, and :*_and_descendents.
 
-The full list of possible synonyms are shown in the spec examples. In 
+The full list of possible synonyms are shown in the spec examples. In
 particular, see "pointcut_spec.rb".
 
 == Version 0.2.0
 
-V0.2.0 changes the parameter list used for advice blocks and adds numerous 
+V0.2.0 changes the parameter list used for advice blocks and adds numerous
 enhancements, robustness improvements and more complete "spec'ing".
 
 Bug fixes:
@@ -325,20 +350,20 @@ none
 Enhancements:
 13402	Support a subclass syntax like AspectJ's "Type+".
 13984	More flexible argument list to the advise block.
-14053	Remove JoinPoint#type, JoinPoint#type=, JoinPoint#object, and 
+14053	Remove JoinPoint#type, JoinPoint#type=, JoinPoint#object, and
   JoinPoint#object=
 14061	Add a control flow mechanism to skipping (sic) intermediate advice
 15164	Deprecate ObjectFinder
 15413	Remove ObjectFinder
 15710	Eliminate redundant public methods in various "finders"
 
-#13402 adds new invocation options to specify types and their descendents 
-  (subclasses or modules that include the specified module(s)) and ancestors. 
-  The latter should be used cautiously as it will include things like Kernel, 
-  Object, and Class!  I used new command options rather than the AspectJ "+" 
-  suffix (and the proposed, but never implemented "-" suffix for ancestors), 
-  because the "+" would be confusing with regular expressions and not in the 
-  spirit of trying to make the pointcut language "easy to read". So, the 
+#13402 adds new invocation options to specify types and their descendents
+  (subclasses or modules that include the specified module(s)) and ancestors.
+  The latter should be used cautiously as it will include things like Kernel,
+  Object, and Class!  I used new command options rather than the AspectJ "+"
+  suffix (and the proposed, but never implemented "-" suffix for ancestors),
+  because the "+" would be confusing with regular expressions and not in the
+  spirit of trying to make the pointcut language "easy to read". So, the
   following are now available:
     :type_and_ancestors
     :types_and_ancestors
@@ -349,39 +374,39 @@ Enhancements:
     :exclude_types_and_ancestors
     :exclude_type_and_descendents
     :exclude_types_and_descendents
-  If you want both the ancestors and descendents, just use both options with 
+  If you want both the ancestors and descendents, just use both options with
   the same value.
 
-#13984 adds the object as the second argument to the advice block parameter 
-  list. This change reflects the fact that the object is often needed, but 
-  calling jp.context.advised_object is a bit tedious. THIS CHANGE BREAKS 
-  BACKWARDS COMPATIBILITY!! An exception is raised if advice has the signature 
+#13984 adds the object as the second argument to the advice block parameter
+  list. This change reflects the fact that the object is often needed, but
+  calling jp.context.advised_object is a bit tedious. THIS CHANGE BREAKS
+  BACKWARDS COMPATIBILITY!! An exception is raised if advice has the signature
   |jp, *args|.
 
-#14061 adds a new method, JoinPoint#invoke_original_join_point, which will 
-  invoke the original method without intermediate advice. If called within 
-  around advice, you can write advice that vetoes all subsequent advice, yet 
-  invokes the original method. Use this technique cautiously, however, since you 
-  may not always know what other advices are involved and what side effects this 
+#14061 adds a new method, JoinPoint#invoke_original_join_point, which will
+  invoke the original method without intermediate advice. If called within
+  around advice, you can write advice that vetoes all subsequent advice, yet
+  invokes the original method. Use this technique cautiously, however, since you
+  may not always know what other advices are involved and what side effects this
   control-flow change might cause.
 
-#15164 and 15413 remove ObjectFinder because it is not used and it requires 
-  ObjectSpace, which has high overhead and won't be enabled, by default, in 
+#15164 and 15413 remove ObjectFinder because it is not used and it requires
+  ObjectSpace, which has high overhead and won't be enabled, by default, in
   JRuby (it will be optional).
 
-#15710 removes redundant methods that were becoming a maintenance issue, in 
-  particular, MethodFinder#find_all_by and TypeFinder#find_by_name. This is a 
-  non-backwards-compatible API change. 
+#15710 removes redundant methods that were becoming a maintenance issue, in
+  particular, MethodFinder#find_all_by and TypeFinder#find_by_name. This is a
+  non-backwards-compatible API change.
 
-Finally, note that I have not yet been able to resolve bug #15202, "Intermittent 
-confusion between classes and objects when invoking advice." I believe this is a 
-very rare occurrence and only likely to ever happen during the "torture-test" of 
+Finally, note that I have not yet been able to resolve bug #15202, "Intermittent
+confusion between classes and objects when invoking advice." I believe this is a
+very rare occurrence and only likely to ever happen during the "torture-test" of
 running the RSpec suite. Please post a comment to Tracker if you encounter it!
 
 == Version 0.1.8
 
-V0.1.7 did not successfully "register" at rubyforge. This releases fixes that 
-problem and also adds several feature enhancements and refactorings. There are 
+V0.1.7 did not successfully "register" at rubyforge. This releases fixes that
+problem and also adds several feature enhancements and refactorings. There are
 no known upgrade issues.
 
 Bug fixes:
@@ -391,14 +416,14 @@ Enhancements:
 13399	Add :exclusion options for methods and types.
 14707	:exclude_ancestor_methods as synonym for :suppress_ancestor_methods
 
-13399 adds new :exclude_(pointcuts|join_points|types|objects|methods|attributes) 
-  options for Aspect.new, and Pointcut.new that make it easier to specify a list 
-  or regular expression for various "items" and then to exclude particular 
-  items, e.g., Aspect.new :around, :types => /nterestingType/, :exclude_types => 
+13399 adds new :exclude_(pointcuts|join_points|types|objects|methods|attributes)
+  options for Aspect.new, and Pointcut.new that make it easier to specify a list
+  or regular expression for various "items" and then to exclude particular
+  items, e.g., Aspect.new :around, :types => /nterestingType/, :exclude_types =>
   UninterestingType ...
 
-The :exclude_ancestor_methods option is now preferred over 
-:suppress_ancestor_methods, since the former is more consistent with the new 
+The :exclude_ancestor_methods option is now preferred over
+:suppress_ancestor_methods, since the former is more consistent with the new
 :exclude_* options.
 
 == Version 0.1.7
@@ -407,43 +432,43 @@ Bug fixes:
 14946	Advice fails when instrumenting methods containing special characters
 15038	Spec for pointcut example variation #1
 15039	Spec for pointcut example variation #2
-15085	Specifying just :attributes for aspects also matches all methods, as if 
+15085	Specifying just :attributes for aspects also matches all methods, as if
   :methods => :all specified
 
 Enhancements:
 13396	Unify internal handling of types vs. objects
 
-15038 and 15039 were bugs in one of the examples (actually in the comments). 
+15038 and 15039 were bugs in one of the examples (actually in the comments).
   However, experimenting with them also revealed the nasty 15085 bug!
 
-I previously handled some special characters in method names, but not all the 
-possible ones, hence 14946. Aquarium should now properly handle any valid Ruby 
+I previously handled some special characters in method names, but not all the
+possible ones, hence 14946. Aquarium should now properly handle any valid Ruby
 method name.
 
 == Version 0.1.6
 
 Bug fixes:
-14353	Advising subclass method that calls super raises exception when method 
+14353	Advising subclass method that calls super raises exception when method
   executed
-14356	Regexps for types must cover the whole name, which is inconsistent with 
+14356	Regexps for types must cover the whole name, which is inconsistent with
   method/attribute regexps
 14384	Design by Contract "extra" does not return correct value "invar" handling
 13410	Fix funky navigation bar on website
 
-14353 was kind of bad, but it's actually a Ruby bug with a good workaround. If 
-  you advised a method that called "super", Ruby would use the wrong method name 
+14353 was kind of bad, but it's actually a Ruby bug with a good workaround. If
+  you advised a method that called "super", Ruby would use the wrong method name
   to lookup the class in the parent. See the bug description for the details.
 
-For 14356, type regular expressions now match on parts of names; they don't have 
-  to match the whole name. The exception is regular expressions with module 
-  separators "::". In this case, it seems to make more sense for the regular 
-  expression to be interpreted as follows: If the expression is /A::B::C::D/, 
-  then for the the outermost types, the expression behaves as /^.*A/, for the 
-  types between two "::", the expressions behave as /^B$/ and /^C$/, and the 
+For 14356, type regular expressions now match on parts of names; they don't have
+  to match the whole name. The exception is regular expressions with module
+  separators "::". In this case, it seems to make more sense for the regular
+  expression to be interpreted as follows: If the expression is /A::B::C::D/,
+  then for the the outermost types, the expression behaves as /^.*A/, for the
+  types between two "::", the expressions behave as /^B$/ and /^C$/, and the
   trailing expression behaves as /D.*$/.
 
-14384 was an easy mistake to make with "around" advice; you have to remember to 
-  return the result of the "join_point.proceed" call, unless you specifically 
+14384 was an easy mistake to make with "around" advice; you have to remember to
+  return the result of the "join_point.proceed" call, unless you specifically
   want to change the returned value! Here are two ways to do it:
 
 	do_something_before(...)
@@ -452,7 +477,7 @@ For 14356, type regular expressions now match on parts of names; they don't have
 	return result
 
   or
-	
+
 	begin
 		do_something_before(...)
 		join_point.proceed
@@ -460,10 +485,10 @@ For 14356, type regular expressions now match on parts of names; they don't have
 		do_something_after(...)
 	end
 
-The latter approach looks "asymmetrical" and it will behave differently if 
-"proceed" raises! However, it eliminates the temporary, if you find that 
+The latter approach looks "asymmetrical" and it will behave differently if
+"proceed" raises! However, it eliminates the temporary, if you find that
 desirable.
- 
+
 Enhancements:
 13407	Pick a better method name for JoinPoint#type, which hides the Module#type
 14385	Pointcut.new should accept a :join_point => jp argument
@@ -476,24 +501,24 @@ For 13407, new attribute methods have been added
 * JoinPoint#target_object 	return the object that the join_point matches.
 * JoinPoint#target_object=	set the object that the join_point matches.
 
-The following, older methods are now deprecated and will be removed in the 0.2.0 
+The following, older methods are now deprecated and will be removed in the 0.2.0
 release (#14053):
 * JoinPoint#type
 * JoinPoint#type=
 * JoinPoint#object
 * JoinPoint#object=
 
-JoinPoint#type method is deprecated because it hides Module#type, which returns 
-the type of the corresponding object. For "symmetry", the other three methods 
-are also now deprecated and they will be removed in a future release. Until 
-then, all will print a warning message to STDOUT. (If you really want the type 
-of what could be a JoinPoint object, you should use #class anyway, as 
+JoinPoint#type method is deprecated because it hides Module#type, which returns
+the type of the corresponding object. For "symmetry", the other three methods
+are also now deprecated and they will be removed in a future release. Until
+then, all will print a warning message to STDOUT. (If you really want the type
+of what could be a JoinPoint object, you should use #class anyway, as
 Module#type is also deprecated!)
 
 == Version 0.1.5
 
 Bug fixes:
-13514	Protected and private methods are made public when advised and left that 
+13514	Protected and private methods are made public when advised and left that
   way when unadvised
 13650	Loading Aquarium interferes with Rails filters
 13864	Bug with negative object_id
@@ -502,23 +527,23 @@ Enhancements:
 13392	Convert examples to specs.
 13463	Support running in JRuby
 
-Fixing 13650 required an API change, which is why I've tagged this release 
-"0.1.5" instead of something like "0.1.1" (and the changes don't seem big enough 
+Fixing 13650 required an API change, which is why I've tagged this release
+"0.1.5" instead of something like "0.1.1" (and the changes don't seem big enough
 to warrant "0.2.0"...).
 
-Previously, requiring "aquarium.rb" in the top-level "lib" directory would 
-implicitly require lib/aquarium/dsl/aspect_dsl.rb, which has Object include the 
-AspectDSL module. This module adds methods like :before and :after to Object. 
-Unfortunately, those methods collide with methods of the same name that Rails 
-adds to Object. It was also a bit presumptuous of me to assume that everyone 
+Previously, requiring "aquarium.rb" in the top-level "lib" directory would
+implicitly require lib/aquarium/dsl/aspect_dsl.rb, which has Object include the
+AspectDSL module. This module adds methods like :before and :after to Object.
+Unfortunately, those methods collide with methods of the same name that Rails
+adds to Object. It was also a bit presumptuous of me to assume that everyone
 wanted those methods on Object ;)
 
-In this release, aspect_dsl.rb is still implicitly included and it still defines 
-the AspectDSL module. Now, however, it does not include the AspectDSL module in 
-Object. Instead, if you want this behavior for all types, you must require the 
-new lib/aquarium/aspects/dsl/object_dsl.rb explicitly. 
+In this release, aspect_dsl.rb is still implicitly included and it still defines
+the AspectDSL module. Now, however, it does not include the AspectDSL module in
+Object. Instead, if you want this behavior for all types, you must require the
+new lib/aquarium/aspects/dsl/object_dsl.rb explicitly.
 
-As an alternative, if you just want the AspectDSL module included selectively in 
+As an alternative, if you just want the AspectDSL module included selectively in
 certain types, then do the following:
 
 	class MyClass   # reopen "MyClass"
@@ -541,12 +566,12 @@ To add the methods as _instance_ methods on individual objects:
 	object.extend(Aquarium::DSL)
 
 
-Note: as discussed at 
-http://practicalruby.blogspot.com/2007/02/reopen-with-moduleeval.html, 
-using "class_eval" or "module_eval" is safer that just reopening a class if 
-you're not sure that "MyClass" has actually been defined yet. However, in our 
-particular case, it probably doesn't matter, as AspectDSL doesn't change 
-anything about the type, like aliasing existing methods. Still, we can't 
+Note: as discussed at
+http://practicalruby.blogspot.com/2007/02/reopen-with-moduleeval.html,
+using "class_eval" or "module_eval" is safer that just reopening a class if
+you're not sure that "MyClass" has actually been defined yet. However, in our
+particular case, it probably doesn't matter, as AspectDSL doesn't change
+anything about the type, like aliasing existing methods. Still, we can't
 guarantee that this won't change in the future.
 
 == Version 0.1.0