[ANN] JRuby 1.4.0RC2 Released

[Thomas E Enebo]

The JRuby community is pleased to announce the release of JRuby 1.4.0RC2

Homepage: http://www.jruby.org/
Download: http://www.jruby.org/download

  The JRuby 1.4.0 release ends a long release cycle in which we
resolved a huge number of issues. The headline feature is a move to
Ruby 1.8.7 support, in line with Rails' own plans to make 1.8.7 the required
minimum version of Ruby.

  We have also gotten more serious about out 1.9 support (--1.9).  The
bigger applications mostly just work at this point, but we still have a
ways to go on 1.9 support.  Please try your 1.9 code in JRuby and help us
firm things up.

  Due to the overwhelming feedback we got from RC1, we have fixed a
number of reported issues (shown below) for RC2.  We expect that this will
be our last RC before 1.4.0 final.

****Please try your apps against 1.4.0RC2 ASAP and report problems****

1.4.0RC2 Highlights:
- Windows Native Launcher
  - Solves many reported bugs with our old .bat scripts

1.4.0RC1 Highlights:
- JRuby is now compatible with 1.8.7 patchlevel 174
- New YAML parser Yecht (bug-for-bug compatible with Syck!)
- Improved Java Integration Support
    - faster
    - new java_method and java_send methods
- Windows installer
- New embedding framework
    - Java 6 scripting out of the box
    - New org.jruby.embed package to replace JavaEmbedUtils
- Many 1.9 mode (--1.9) fixes
    - updated stdlib to 1.9.2preview2
    - Rubygems now working
    - Rails 2.3.3 simple apps working (2.3.4 has 1.9 bug)
- rubygems 1.3.5
- Much better error reporting
    - shorter backtraces with less Java
    - Ruby-compatible syntax error reporting
- 232 bugs fixed since 1.3.1

  We want to give a couple of special acknowledgments for some contributors
who went the extra mile this release cycle.

David Calavera has done a great job wrestling with Rubyspec failures and
coming up with patches.

Yoko Harada has totally revamped our embedding APIs and created a new API
called Red Bridge.  This should address all of the reported embedding issues.

Issues fixed since 1.4.0RC1:
JRUBY-4036 Regression: Instead of java-based ThreadLibrary a
pure-ruby thread.rb from stdlib is used
JRUBY-4064 Yecht regression: require'ing rubygems now makes nil,
true, false all tainted
JRUBY-4051 Recieve error "\jruby-1.4.0RC1\bin\) was unexpected at
this time" when i try use jruby or jirb
JRUBY-4063 memory stomping error with debug JVM on Solaris
JRUBY-4042 [windows] jruby.bat without parameters fails
JRUBY-4055 [windows] jruby.bat can't handle caret sign (^) and
question mark (?) in command line
JRUBY-4058 [windows] jruby.bat eats % signs
JRUBY-4056 unsigned values values should be stuffed into Java signed
values like Java does
JRUBY-4039 Implicit conversion to List<Long> results in a List<Byte>
or List<Integer>
JRUBY-4079 rcov fails with NullPointerException with JRuby 1.4.0RC1.
Has worked with all JRuby versions since 2008.
JRUBY-4041 tool/nailgun/configure is not executable
JRUBY-4069 [windows] jruby.jar is not in bootclasspath, JRuby starts
much slower then it colud
JRUBY-4071 SystemCallError.new does not create an Errno instance
JRUBY-4089 Mangled Joda-Time package breaks public APIs that return
or consume Joda-Time objects
JRUBY-4045 [windows] jruby -e "p $LOAD_PATH" fails (regression)
JRUBY-4066 ruby subclass of java public void method cannot return
numeric values
JRUBY-4043 could not coerce BigDecimal to class java.lang.Object (TypeError)
JRUBY-4053 ActiveRecord AssociationCollection#== method is returning
false on equal results
JRUBY-4094 variadic FFI functions with struct-by-value return types are broken
JRUBY-4062 RUBYOPT=-rubygems causes various problems for JRuby
JRUBY-4060 gem command doesn't work when RUBYOPT set
JRUBY-4038 nil implicitly casts to java.lang.Integer instead of
java.lang.Long during method dispatch
JRUBY-4095 FFI functions that return a struct-by-value cannot accept
non-primitive parameters
JRUBY-4111 Solaris 32bit stat struct too large
JRUBY-4108 LoadService::SuffixType too private now for LoadService magic
JRUBY-4131 Nailgun doesn't work on Windows 2000
JRUBY-4129 gem install can not work when the gem name contails '_p'

Issues Fixed for 1.4.0RC1:
Key        Summary
JRUBY-382     Better error output from parser
JRUBY-937     JRuby does not honor RUBY_OPTS or RUBYOPT
JRUBY-2232     UNIXSocket doesn't work on Solaris
JRUBY-2324     AST Node positions and content incorrect when passing
down Multibyte content to lexer
JRUBY-2349     Bug in constant lookup on non Module or Class
JRUBY-2398     We should strive to implement the 1.8.7 improvements
JRUBY-2419     Argument application within block is not parsed
correctly (for Range literals)
JRUBY-2525     Dir.glob ordering inconsistent with Matzruby
JRUBY-2587     [1.8.7] Hash.[] should call to_ary if only one argument
is provided
JRUBY-2604     [1.8.7] Implement Binding#eval method
JRUBY-2626     Many rubyspec failures for SortedSet
JRUBY-2654     File pty.rb is not part of the JRuby distro, but is
part of the Ruby standard lib
JRUBY-2666     Issues with Socket#getaddrinfo
JRUBY-2694     [1.8.7] Ability to specify suffix/extension for Tempfile
JRUBY-2866     Set .include? method behavior differs between JRuby
1.1.3 and Ruby 1.8.7
JRUBY-3123     Java primate arrays no longer respond to 'first'
(Array) from ruby
JRUBY-3150     Execution variable $: does not include on the taint
check level > 1
JRUBY-3154     Net::HTTP.read_timeout is ignored, timeout is intinite
instead (leading to massive accumulation of sockets and reading
threads)
JRUBY-3159     -e option does not work with iterator blocks
JRUBY-3165     Add stream/reader based methods to
JavaEmbedUtils.RuntimeAdapter as convenience methods
JRUBY-3215     BSF does not retain local scope between eval calls
JRUBY-3240     YAML: active_support dump of BigDecimal differs from MRI
JRUBY-3263     imcompatibility with Ruby-1.8.6 in YAML parsing
JRUBY-3266     undefined method 'add_ruby_type' for YAML:Module
JRUBY-3297     Array.new with (size, object=nil) returns the value
passed to break
JRUBY-3306     --1.9 mode doesn't know define_singleton_method
JRUBY-3319     Date spec failures
JRUBY-3321     matrix rubyspec failures
JRUBY-3323     rational rubyspec failures
JRUBY-3327     Possible YAML parsing error
JRUBY-3352     JNA-based FFI fails to load for specs on Windows
JRUBY-3368     java.util.List[range] is broken
JRUBY-3380     New 1.9 Enumerator methods
JRUBY-3389     RubySpec: Hash#update raises a TypeError if called on a
non-empty, frozen instance
JRUBY-3403     Socket close_read and close_write should use
appropriate methods on java.net.Socket
JRUBY-3412     YAML.dump loses the timezone when dumping Time instances
JRUBY-3419     JRuby Fails (MRI 1.8.6 works) in Sinatra "helpers" when
no arguments
JRUBY-3433     Encoding.default_internal and Encoding.default_external missing
JRUBY-3442     1.9 allows trailing commas in method invocations
JRUBY-3453     Make import not complain if Java already has already
imported the specific Java class
JRUBY-3473     RubySpec: OpenStruct frozen failures
JRUBY-3478     Dir.new does not work for directories inside jar file
JRUBY-3499     File.expand_path (from inside a jar) uses %20 instead
of space characters
JRUBY-3505     Can't install hpricot (0.7) gem
JRUBY-3518     JRUBY-1786
yaml_initialize not called for subclasses of standard types
JRUBY-3545     ArrayIndexOutOfBoundsException when opening
Bio::FlatFiles (bioruby gem)
JRUBY-3546     ~/.hoerc results in parsing error
JRUBY-3558     RubySpec: Defining instance methods on an eigenclass
define public methods
JRUBY-3559     the integratated joda time version should be in a
distinct package hierarchy
JRUBY-3575     File permission errors installing gems in 1.9 mode
JRUBY-3576     utf-8 string becomes macroman on os x
JRUBY-3577     File and FileTest methods should use to_path to coerce
non-string parameters
JRUBY-3591     Hexadecimal constants do not have the same equal?
behaviour between JRuby and CRuby
JRUBY-3592     Regression test failure (test_grpowned)
JRUBY-3605     [1.9] IO.initialize missing option hash parsing
JRUBY-3606     [1.9] IO.set_encoding is stubbed out
JRUBY-3607     [1.9] IO.external_encoding stubbed out
JRUBY-3611     RubySpec: "StringIO#sysread when passed no arguments" failures
JRUBY-3612     RubySpec: Array#to_s returns '[...]' for the embedded
array in a recursive array
JRUBY-3627     Parsing related methods and class/interdace should have
ManyVarsDynamicScope to share local variables
JRUBY-3653     RubySpec: Hash failures "computes equality for complex
recursive hashes"
JRUBY-3654     RubySpec: Array failures relating to recursive arrays
JRUBY-3662     JavaEmbedUtils should have methods to set/reset Standard I/O
JRUBY-3682     Hashcodes for Hash objects break compatability with MRI
JRUBY-3687     Bundle JSR 223 support in JRuby
JRUBY-3688     IO.read fails to read files in /proc filesystem on Linux
JRUBY-3700     test/externals/ruby1.8/yaml/test_yaml failures
JRUBY-3702     The method methods
JRUBY-3703     TestCommandLineSwitches fails on Windows
JRUBY-3705     Rubyspec: behavior of $: under safe > 1
JRUBY-3706     Rubyspec: method arity behavior
JRUBY-3707     Rubyspec: Process egid behavior
JRUBY-3718     FFI: put_pointer and put_array_of_pointer fails to send
:to_ptr to their argument
JRUBY-3721     sub-optimal IO#read after IO#getc
JRUBY-3722     block parameter problem
JRUBY-3725     Fix for JRUBY-3563 introduces regression in Socket#getnameinfo
JRUBY-3731     --debug causes problems. Methods are missing.
JRUBY-3736     JRuby won't load a .jar if a .so is found too.
JRUBY-3738     jruby 1.3.0 windows vista issues: doesn't output
helpful error if JAVA_HOME undefined, and JAVA_HOME path defined in
quotes not supported
JRUBY-3740     Thread#wakeup not working
JRUBY-3743     Timeout no longer supports sub-second timeouts
JRUBY-3744     Ruby BigDecimal vulnerability seems to affect JRuby as well
JRUBY-3745     RubySpec: Hash#rehash causes a RuntimeError to be
raised if called inside an iterator block
JRUBY-3747     String#gsub! with pattern and block raises a
RuntimeError when self is frozen
JRUBY-3748     RubySpec: Kernel.Integer raises an ArgumentError if
there is a space between the + and number
JRUBY-3750     JRUBY_OPTS="--1.9" results in error "jruby: unknown option --1.9"
JRUBY-3751     YAML.load fails for Struct objects with non-standard initialize
JRUBY-3757     String.split is broken
JRUBY-3759     syslog not available for JRuby --1.9
JRUBY-3761     FileUtils.chown_R produces ArgumentError when the
directory contains a symlink
JRUBY-3766     update .classpath for building jruby with eclipse
JRUBY-3773     CLONE -custom YAML serialization ignored by JRuby
JRUBY-3774     >>define_method :name do | &blk | << ignores
JRUBY-3776     The install-gems goal throws exceptions
JRUBY-3777     Exception on Solaris loading ruby gems
JRUBY-3778     clone of stderr and stdout not working
JRUBY-3782     BSF interface - apply nonfunctional with parameters
JRUBY-3783     Timeout.timeout doesn't raise given exception
JRUBY-3784     More than 2G memory required for jruby -e 'buf =
IO.read("/tmp/1GB.txt"); p buf.size'
JRUBY-3788     argument population of blocks broke in at least one
case somewhere between 1.2 and 1.3
JRUBY-3799     Bug in native timeout module causes hang when waiting on IO
JRUBY-3800     CLASSPATH environment variable should be ignored if
-J-cp or -J-classpath are provided
JRUBY-3806     Encoding problems with File.expand_path after
JRUBY-3576 fix for Macroman issue
JRUBY-3807     jgem --1.9 broken
JRUBY-3808     Gem redclothcoderay doesn't installs under jruby
JRUBY-3809     require 'tmpdir' fails when embedded in a java app that
has messed with the classpath
JRUBY-3810     StringIO#truncate problem
JRUBY-3811     require 'Complex' fails when running in Ruby 1.9 mode
JRUBY-3815     Time#strftime breaks with multibyte characters, affects
I18n#localize
JRUBY-3817     Timeouts in Net::HTTP raise Timeout::ExitException
instead of Timeout::Error for Ruby 1.8
JRUBY-3820     timeout library should throw an anonymous exception
type while unrolling stack
JRUBY-3823     Wrong property in netbeans-ant.xml
JRUBY-3827     WEBrick binds only to localhost, not to requested bind addresses
JRUBY-3828     JavaUtil.convertRubyToJava fails to convert objects in
seemingly random cases
JRUBY-3832     FFI: apparent memory leak when using callbacks
JRUBY-3835     FFI AbstractMemory#get_array_of_uint64 is missing
JRUBY-3836     backquotes can raise IO Error (closed stream)
JRUBY-3838     Exceptions coming out of eval no longer have line numbers
JRUBY-3849     Resolv.getaddress is broken on OSX 1.5
JRUBY-3850     Hash#select returns an Array of entries for which block
is true FAILED
JRUBY-3851     Hash#update raises a TypeError if called on a
non-empty, frozen instance ERROR
JRUBY-3852     Hash#eql? spec failures
JRUBY-3853     Two Enumerable spec bugs for 1.8.7 (one related to break)
JRUBY-3854     Five spec failures in ConditionVariable after merge
JRUBY-3856     YAML.load is broken when loading structs with
initialize method with arguments
JRUBY-3858     NPE in RubyRegexp.java with --1.9
JRUBY-3859     VariadicInvoker missing a BOOL case
JRUBY-3860     Rational(3,-4).to_i should return -1
JRUBY-3861     Shared "box" in DefaultMethod breaking visibility
re-checks in some cases
JRUBY-3862     Socket.pack_sockaddr_in is broken on linux
JRUBY-3863     Support boolean arguments to and return values from ffi callbacks
JRUBY-3867     Nailgun not supported in Windows bat files
JRUBY-3869     'puts []' prints a newline
JRUBY-3870     Object#singleton_methods(true) doesn't include inherited methods
JRUBY-3871     strftime doesn't support %h
JRUBY-3873     StringIndexOutOfBoundsException if the last character
in format is '%' in Time.strftime
JRUBY-3874     it would be nice if LoadService could log where it
searches and what it finds
JRUBY-3875     TypeError when providing nil as the third parameter to
TCPSocket.new/TCPSocket.open
JRUBY-3876     JRuby crashes on using the three-parameter version of
TCPSocket.open
JRUBY-3878     String representation of arrays and symbols does not
match ruby 1.9
JRUBY-3879     Array#pack('d') is incorrect
JRUBY-3882     Java raised exceptions fail to list JRuby caller method
in the backtrace
JRUBY-3884     Rubyspec failures: Dir#inspect includes the directory name
JRUBY-3885     Rubyspec failures: Enumerable#none?
JRUBY-3889     java errors are hidden when an instance of a class is
created and we used the method include_package to load the specific
java package
JRUBY-3890     NativeExceptions have ugly stack traces
JRUBY-3891     JavaSupport.handleNativeException doesn't seem handle
superclasses properly
JRUBY-3892     Running rake rdoc from within jruby-complete.jar
results in "undefined method `key' for nil:NilClass"
JRUBY-3893     Can't require yaml from jruby-complete.jar at head
JRUBY-3894     Problems with __FILE__ and file loaded from a jar
JRUBY-3896     '--ng' in JRUBY_OPTS is not passed correctly
JRUBY-3897     Dir::entries doesn't work within a jar
JRUBY-3907     Time.strftime allows tokens deriving from GNU C library
JRUBY-3908     Addendum to JRUBY-3896
JRUBY-3910     Java List to Ruby Array conversion ~100% slower than
primitive array conversion
JRUBY-3911     hpricot is busted when truncating multibyte strings
JRUBY-3912     FFI - Some Ruby-FFI specs fail on JRuby
JRUBY-3913     Date.strftime bug fixes (%u, %w, %s) and addition of %G
JRUBY-3914     In 1.9 mode, Time.now == nil should return false, not nil
JRUBY-3916     Clean up nailgun directory a little
JRUBY-3922     File.basename doesn't cope with UTF8 strings
JRUBY-3927     Jar complete doesn't compile
JRUBY-3929     checksums for new versions please
JRUBY-3932     Set == broken in 1.9
JRUBY-3933     [19] IO.gets missing additional limit field
JRUBY-3936     Math.gamma fails on certain values when it shouldn't
JRUBY-3937     Socket::getaddrinfo() returns incorrect data on solaris
JRUBY-3941     [19] socket.do_not_reverse_lookup is unimplemented
JRUBY-3942     [1.9] File.new should accept encoding options within
the mode parameter and a hash with options as the third parameter
JRUBY-3945     [1.9] IO and file initialization should ignore internal
encoding when is equal to external encoding
JRUBY-3946     [1.9] Encoding.default_external should get encoding
from default charset when is nil
JRUBY-3947     [1.9] Domain errors on Math.log(0), Math.log2(0),
Math.log10(0), Math.atanh(1), Math.atanh(-1)
JRUBY-3951     TMail fails to install without modifications mkmf.rb
JRUBY-3952     Update pom.xml to reflect move from SVN to Git.
JRUBY-3953     jruby.bat doesn't run on Windows 7
JRUBY-3954     JRuby operating differently from MRI with andand gem
JRUBY-3956     Can't do Enumerator#next in --1.9? Not implemented yet?
JRUBY-3967     remove some duplicated tests that have been added to rubyspecs
JRUBY-3968     YAML.load produces: allocator undefined for Symbol (TypeError)
JRUBY-3969     LICENSE.RUBY is missing in bin and source distributive
for the JRuby 1.3.1 and 1.3
JRUBY-3970     jruby.bat errors out with unexpected characters
JRUBY-3974     In-proc execution loses almost all parent-process config values
JRUBY-3977     "require" won't load jars with ".so" anywhere in the name
JRUBY-3981     Cannot connect to ActiveRecord database
JRUBY-3984     [1.9] Array improvements to solve several specs
JRUBY-3985     [1.9] ARGF improvements to solve several specs
JRUBY-3987     [1.9] Bignum improvements to pass some ruby specs
JRUBY-3988     Fixnums quietly lop off bits when calling methods of
lesser precision than long
JRUBY-3989     nil can't be passed for numeric Java types in some
cases, even though nil.to_i treats it as 0
JRUBY-3990     Fixnums are inconsistently converted when target type
is java.lang.Object
JRUBY-3991     [1.9] Comparable improvements to pass ruby specs
JRUBY-3993     [1.9] Dir improvements to pass some ruby specs
JRUBY-3995     ant bails with unrecognized -d32 option on Windows
JRUBY-3997     IBM JDK does not support some String.format specifiers
JRUBY-3998     Path separators differ between jruby and jruby --ng on Windows
JRUBY-3999     [PATCH] Improved Sorting Algorithm
JRUBY-4000     [1.9] Fixnum improvements to solve ruby specs
JRUBY-4001     [PATCH] Fixed float rubyspecs
JRUBY-4003     NegativeArraySizeException in RubyHash.keys
JRUBY-4004     jrubyc should return a non-zero return code if compilation fails
JRUBY-4011     IO.binmode should raise an IOError on closed stream
JRUBY-4017     [1.8] Method#name and UnboundMethod#name should return
strings, not symbols
JRUBY-4019     Build issues when changes from rev 2eef39 are included
JRUBY-4021     [1.9] undefined methods Hash.compare_by_identity and
Hash.compare_by_identity?
JRUBY-4022     9 rubyspec failures for SecureRandom


--
blog: http://blog.enebo.com       twitter: tom_enebo
mail: tom.enebo@...

---------------------------------------------------------------------
To unsubscribe from this list, please visit:

    http://xircles.codehaus.org/manage_email

[Trejkaz-3]

On Thu, Oct 22, 2009 at 8:47 AM, Thomas E Enebo <tom.enebo@...> wrote:
> The JRuby community is pleased to announce the release of JRuby 1.4.0RC2
>
> Homepage: http://www.jruby.org/
> Download: http://www.jruby.org/download

Might want to update the link text on this page.  The files are RC2
but the links all say RC1 still. :-)

TX

---------------------------------------------------------------------
To unsubscribe from this list, please visit:

    http://xircles.codehaus.org/manage_email

[Thomas E Enebo]

On Wed, Oct 21, 2009 at 5:25 PM, Trejkaz <trejkaz@...> wrote:

> On Thu, Oct 22, 2009 at 8:47 AM, Thomas E Enebo <tom.enebo@...> wrote:
>> The JRuby community is pleased to announce the release of JRuby 1.4.0RC2
>>
>> Homepage: http://www.jruby.org/
>> Download: http://www.jruby.org/download
>
> Might want to update the link text on this page.  The files are RC2
> but the links all say RC1 still. :-)
>
> TX

Updated.  Thanks for the report.

-Tom

--
blog: http://blog.enebo.com       twitter: tom_enebo
mail: tom.enebo@...

---------------------------------------------------------------------
To unsubscribe from this list, please visit:

    http://xircles.codehaus.org/manage_email

[Jon M]

> The JRuby community is pleased to announce the release of JRuby 1.4.0RC2
>
> Homepage: http://www.jruby.org/
> Download: http://www.jruby.org/download

Nice new site.

Maybe set the mime type for the MD5 and SHA1 hashes to text/plain (similar to http://maven.apache.org/download.html and http://www.scala-lang.org/downloads but also include the filename) so they can be both viewed in a tab or downloaded depending on which you prefer?

Jon

---------------------------------------------------------------------
To unsubscribe from this list, please visit:

    http://xircles.codehaus.org/manage_email

[Roger Pack-5]

> - Windows installer

Yea!
When I install using the installer, it says "install4j wizard
The Java VM could not be started"

then a few seconds later, up pops the normal, expected setup window, and
it installs normally.

Thanks!
-r
--
Posted via http://www.ruby-forum.com/.

---------------------------------------------------------------------
To unsubscribe from this list, please visit:

    http://xircles.codehaus.org/manage_email

[Thomas E Enebo]

Our backend stuff is being kajiggered right now so until we get that
straightened out I captured this as a bug:
http://jira.codehaus.org/browse/JRUBY-4143.  Thanks for the report.
It would make it easier to compare hashes.

-Tom

On Thu, Oct 22, 2009 at 8:47 AM, Jon <jon.forums@...> wrote:

>> The JRuby community is pleased to announce the release of JRuby 1.4.0RC2
>>
>> Homepage: http://www.jruby.org/
>> Download: http://www.jruby.org/download
>
> Nice new site.
>
> Maybe set the mime type for the MD5 and SHA1 hashes to text/plain (similar to http://maven.apache.org/download.html and http://www.scala-lang.org/downloads but also include the filename) so they can be both viewed in a tab or downloaded depending on which you prefer?
>
> Jon
>
> ---------------------------------------------------------------------
> To unsubscribe from this list, please visit:
>
>    http://xircles.codehaus.org/manage_email
>
>
>

--
blog: http://blog.enebo.com       twitter: tom_enebo
mail: tom.enebo@...

---------------------------------------------------------------------
To unsubscribe from this list, please visit:

    http://xircles.codehaus.org/manage_email

[Thomas E Enebo]

Weird.  Anyone else see that or at least know what is going on?

-Tom

On Thu, Oct 22, 2009 at 10:45 AM, Roger Pack <lists@...> wrote:

>
>> - Windows installer
>
> Yea!
> When I install using the installer, it says "install4j wizard
> The Java VM could not be started"
>
> then a few seconds later, up pops the normal, expected setup window, and
> it installs normally.
>
> Thanks!
> -r
> --
> Posted via http://www.ruby-forum.com/.
>
> ---------------------------------------------------------------------
> To unsubscribe from this list, please visit:
>
>    http://xircles.codehaus.org/manage_email
>
>
>

--
blog: http://blog.enebo.com       twitter: tom_enebo
mail: tom.enebo@...

---------------------------------------------------------------------
To unsubscribe from this list, please visit:

    http://xircles.codehaus.org/manage_email

[Shot (Piotr Szotkowski)]

Thomas E Enebo:

> We have also gotten more serious about out 1.9 support (--1.9). The
> bigger applications mostly just work at this point, but we still have
> a ways to go on 1.9 support. Please try your 1.9 code in JRuby and
> help us firm things up.

Ok, I have a problem with JRuby 1.4 not
seeing gems in --1.9 compatibility mode:


shot@devielle:~/PhD/art-decomp$ jruby --1.9 -S rake spec
(in /home/shot/PhD/art-decomp)
rake aborted!
no such file to load -- spec/rake/spectask
/home/shot/PhD/art-decomp/Rakefile:1:in `require'
(See full trace by running task with --trace)
shot@devielle:~/PhD/art-decomp$ jruby --1.9 -S gem list

*** LOCAL GEMS ***

backports (local)
diff-lcs (local)
rake (local)
rspec (local)
sequel (local)
sources (local)
trollop (local)
shot@devielle:~/PhD/art-decomp$ jruby --1.9 -v
jruby 1.4.0RC2 (ruby 1.9.2dev trunk 24787) (2009-10-21 7e77f32) (Java HotSpot(TM) 64-Bit Server VM 1.6.0_16) [amd64-java]
shot@devielle:~/PhD/art-decomp$


My $GEM_HOME is /home/shot/.gem/jruby/1.8 and JRuby in --1.8
compatibility mode sees the gems there (unfortunately it breaks,
because the backports gem – which I use for 1.8 compatibility of
my code – uses Process.exec, which JRuby doesn’t handle):


shot@devielle:~/PhD/art-decomp$ jruby --1.8 -S rake spec
(in /home/shot/PhD/art-decomp)
/home/shot/.gem/jruby/1.8/gems/backports-1.10.2/lib/backports/1.8.7/process.rb:31:in `require': undefined method 'exec' for module 'Process' (NameError)
  from /home/shot/.rvm/jruby-1.3.1/lib/ruby/site_ruby/1.8/rubygems/custom_require.rb:31:in `require'
  from /home/shot/.gem/jruby/1.8/gems/backports-1.10.2/lib/backports/tools.rb:10:in `require_relative'
  from /home/shot/.gem/jruby/1.8/gems/backports-1.10.2/lib/backports/1.8.7.rb:3
  from /home/shot/.gem/jruby/1.8/gems/backports-1.10.2/lib/backports/1.8.7.rb:2:in `each'
  from /home/shot/.gem/jruby/1.8/gems/backports-1.10.2/lib/backports/1.8.7.rb:2
  from /home/shot/.gem/jruby/1.8/gems/backports-1.10.2/lib/backports/1.8.7.rb:31:in `require'
  from /home/shot/.rvm/jruby-1.3.1/lib/ruby/site_ruby/1.8/rubygems/custom_require.rb:31:in `require'
  from /home/shot/.gem/jruby/1.8/gems/backports-1.10.2/lib/backports/tools.rb:10:in `require_relative'
   ... 19 levels...
  from /home/shot/.gem/jruby/1.8/gems/rspec-1.2.9/lib/spec/runner.rb:51:in `options'
  from /home/shot/.gem/jruby/1.8/gems/rspec-1.2.9/bin/spec:5:in `run'
  from /home/shot/.gem/jruby/1.8/gems/rspec-1.2.9/bin/spec:5
rake aborted!
Command /home/shot/.rvm/jruby-1.3.1/bin/jruby -I"lib"  "/home/shot/.gem/jruby/1.8/gems/rspec-1.2.9/bin/spec" "spec/set_spec.rb" "spec/string_spec.rb" "spec/enumerable_spec.rb" "spec/file_spec.rb" "spec/integer_spec.rb" "spec/art-decomp/graph_spec.rb" "spec/art-decomp/logging_spec.rb" "spec/art-decomp/arch_spec.rb" "spec/art-decomp/fsm_spec.rb" "spec/art-decomp/sep_spec.rb" "spec/art-decomp/bipainter_spec.rb" "spec/art-decomp/blanket_spec.rb" "spec/art-decomp/kiss_spec.rb" "spec/art-decomp/executable_spec.rb" "spec/art-decomp/decomposer_spec.rb" "spec/art-decomp/decomposition_spec.rb" "spec/art-decomp/b_spec.rb" "spec/art-decomp/qv_generator/graph_colouring_spec.rb" "spec/art-decomp/qv_generator/bipainting_spec.rb" "spec/art-decomp/qv_generator/graph_merging_spec.rb" "spec/art-decomp/qu_generator/edge_labels_spec.rb" "spec/art-decomp/qu_generator/block_table_spec.rb" "spec/art-decomp/uv_generator/relevance_spec.rb" "spec/art-decomp/uv_generator/braindead_spec.rb" --options spec/spec.opts failed

(See full trace by running task with --trace)
shot@devielle:~/PhD/art-decomp$ jruby --1.8 -S gem list

*** LOCAL GEMS ***

backports (1.10.2)
diff-lcs (1.1.2)
rake (0.8.7)
rspec (1.2.9, 1.2.6)
sequel (3.5.0)
sources (0.0.1)
trollop (1.15)
shot@devielle:~/PhD/art-decomp$ jruby --1.8 -v
jruby 1.4.0RC2 (ruby 1.8.7 patchlevel 174) (2009-10-21 7e77f32) (Java HotSpot(TM) 64-Bit Server VM 1.6.0_16) [amd64-java]
shot@devielle:~/PhD/art-decomp$


You can get the code I’m trying to rake spec at
http://github.com/Chastell/art-decomp (master is
for Ruby 1.9, ruby-1.8-compat is for Ruby 1.8.7).
Any help with the above will be most appreciated!


— Shot
--
Anyone who finds the MS Office ribbon intuitive is
not a member of the human species as I understand it.
                                     [William Morgan]

[Thomas E Enebo]

Hmmm,  I think we will need to try and get some more information on
this one.   Before we go any further trying to debug this, I noticed
one thing which looks wrong below.  You have two lines like:

from /home/shot/.rvm/jruby-1.3.1/lib/ruby/site_ruby/1.8/rubygems/custom_require.rb:31:in
`require'

It looks like you are loading at least some stuff from an older
version of jruby (1.3.1)?  I have never used rvm so perhaps this is
ok, but it looks wrong at first glance.

Also if the rvm thing ends up being benign, then can you also do a
--trace on the 1.9 run (jruby --1.9 -S rake spec --trace).   Getting a
reduced test case (like a minimal Rakefile with minimum specs showing
the loading problem) would be super helpful also.

-Tom

On Sat, Oct 24, 2009 at 8:56 AM, Shot (Piotr Szotkowski) <shot@...> wrote:

> Thomas E Enebo:
>
>> We have also gotten more serious about out 1.9 support (--1.9). The
>> bigger applications mostly just work at this point, but we still have
>> a ways to go on 1.9 support. Please try your 1.9 code in JRuby and
>> help us firm things up.
>
> Ok, I have a problem with JRuby 1.4 not
> seeing gems in --1.9 compatibility mode:
>
>
> shot@devielle:~/PhD/art-decomp$ jruby --1.9 -S rake spec
> (in /home/shot/PhD/art-decomp)
> rake aborted!
> no such file to load -- spec/rake/spectask
> /home/shot/PhD/art-decomp/Rakefile:1:in `require'
> (See full trace by running task with --trace)
> shot@devielle:~/PhD/art-decomp$ jruby --1.9 -S gem list
>
> *** LOCAL GEMS ***
>
> backports (local)
> diff-lcs (local)
> rake (local)
> rspec (local)
> sequel (local)
> sources (local)
> trollop (local)
> shot@devielle:~/PhD/art-decomp$ jruby --1.9 -v
> jruby 1.4.0RC2 (ruby 1.9.2dev trunk 24787) (2009-10-21 7e77f32) (Java HotSpot(TM) 64-Bit Server VM 1.6.0_16) [amd64-java]
> shot@devielle:~/PhD/art-decomp$
>
>
> My $GEM_HOME is /home/shot/.gem/jruby/1.8 and JRuby in --1.8
> compatibility mode sees the gems there (unfortunately it breaks,
> because the backports gem – which I use for 1.8 compatibility of
> my code – uses Process.exec, which JRuby doesn’t handle):
>
>
> shot@devielle:~/PhD/art-decomp$ jruby --1.8 -S rake spec
> (in /home/shot/PhD/art-decomp)
> /home/shot/.gem/jruby/1.8/gems/backports-1.10.2/lib/backports/1.8.7/process.rb:31:in `require': undefined method 'exec' for module 'Process' (NameError)
>  from /home/shot/.rvm/jruby-1.3.1/lib/ruby/site_ruby/1.8/rubygems/custom_require.rb:31:in `require'
>  from /home/shot/.gem/jruby/1.8/gems/backports-1.10.2/lib/backports/tools.rb:10:in `require_relative'
>  from /home/shot/.gem/jruby/1.8/gems/backports-1.10.2/lib/backports/1.8.7.rb:3
>  from /home/shot/.gem/jruby/1.8/gems/backports-1.10.2/lib/backports/1.8.7.rb:2:in `each'
>  from /home/shot/.gem/jruby/1.8/gems/backports-1.10.2/lib/backports/1.8.7.rb:2
>  from /home/shot/.gem/jruby/1.8/gems/backports-1.10.2/lib/backports/1.8.7.rb:31:in `require'
>  from /home/shot/.rvm/jruby-1.3.1/lib/ruby/site_ruby/1.8/rubygems/custom_require.rb:31:in `require'
>  from /home/shot/.gem/jruby/1.8/gems/backports-1.10.2/lib/backports/tools.rb:10:in `require_relative'
>   ... 19 levels...
>  from /home/shot/.gem/jruby/1.8/gems/rspec-1.2.9/lib/spec/runner.rb:51:in `options'
>  from /home/shot/.gem/jruby/1.8/gems/rspec-1.2.9/bin/spec:5:in `run'
>  from /home/shot/.gem/jruby/1.8/gems/rspec-1.2.9/bin/spec:5
> rake aborted!
> Command /home/shot/.rvm/jruby-1.3.1/bin/jruby -I"lib"  "/home/shot/.gem/jruby/1.8/gems/rspec-1.2.9/bin/spec" "spec/set_spec.rb" "spec/string_spec.rb" "spec/enumerable_spec.rb" "spec/file_spec.rb" "spec/integer_spec.rb" "spec/art-decomp/graph_spec.rb" "spec/art-decomp/logging_spec.rb" "spec/art-decomp/arch_spec.rb" "spec/art-decomp/fsm_spec.rb" "spec/art-decomp/sep_spec.rb" "spec/art-decomp/bipainter_spec.rb" "spec/art-decomp/blanket_spec.rb" "spec/art-decomp/kiss_spec.rb" "spec/art-decomp/executable_spec.rb" "spec/art-decomp/decomposer_spec.rb" "spec/art-decomp/decomposition_spec.rb" "spec/art-decomp/b_spec.rb" "spec/art-decomp/qv_generator/graph_colouring_spec.rb" "spec/art-decomp/qv_generator/bipainting_spec.rb" "spec/art-decomp/qv_generator/graph_merging_spec.rb" "spec/art-decomp/qu_generator/edge_labels_spec.rb" "spec/art-decomp/qu_generator/block_table_spec.rb" "spec/art-decomp/uv_generator/relevance_spec.rb" "spec/art-decomp/uv_generator/braindead_spec.rb" --options spec/spec.opts failed
>
> (See full trace by running task with --trace)
> shot@devielle:~/PhD/art-decomp$ jruby --1.8 -S gem list
>
> *** LOCAL GEMS ***
>
> backports (1.10.2)
> diff-lcs (1.1.2)
> rake (0.8.7)
> rspec (1.2.9, 1.2.6)
> sequel (3.5.0)
> sources (0.0.1)
> trollop (1.15)
> shot@devielle:~/PhD/art-decomp$ jruby --1.8 -v
> jruby 1.4.0RC2 (ruby 1.8.7 patchlevel 174) (2009-10-21 7e77f32) (Java HotSpot(TM) 64-Bit Server VM 1.6.0_16) [amd64-java]
> shot@devielle:~/PhD/art-decomp$
>
>
> You can get the code I’m trying to rake spec at
> http://github.com/Chastell/art-decomp (master is
> for Ruby 1.9, ruby-1.8-compat is for Ruby 1.8.7).
> Any help with the above will be most appreciated!
>
>
> — Shot
> --
> Anyone who finds the MS Office ribbon intuitive is
> not a member of the human species as I understand it.
>                                     [William Morgan]
>
> -----BEGIN PGP SIGNATURE-----
> Version: GnuPG v1.4.9 (GNU/Linux)
>
> iEYEARECAAYFAkrjB5gACgkQi/mCfdEo8UpRQACfZ3vF7wdFMtg0SqLLU43V/1MB
> OSIAoMB4zcjZoyxu+CIfK1PKNMc6aTLQ
> =ZjXS
> -----END PGP SIGNATURE-----
>
>

--
blog: http://blog.enebo.com       twitter: tom_enebo
mail: tom.enebo@...

---------------------------------------------------------------------
To unsubscribe from this list, please visit:

    http://xircles.codehaus.org/manage_email

[Shot (Piotr Szotkowski)]

First: Thanks a lot for your reply! Apologies
for the ambiguities, hopefully clarified below.

Thomas E Enebo:

> Hmmm, I think we will need to try and get some more information on
> this one. Before we go any further trying to debug this, I noticed
> one thing which looks wrong below. You have two lines like:

> from /home/shot/.rvm/jruby-1.3.1/lib/ruby/site_ruby/1.8/rubygems/custom_require.rb:31:in `require'

> It looks like you are loading at least some stuff from an older
> version of jruby (1.3.1)? I have never used rvm so perhaps this
> is ok, but it looks wrong at first glance.

Argh, I’m sorry, I really should’ve clarified this: I’m using an old
version of rvm, because never ones break for me. I simply deleted the
old JRuby 1.3.1 files and put JRuby 1.4.0RC2 there. I also had this
problem before 1.4.RC1 with a JRuby built from trunk.

> Also if the rvm thing ends up being benign, then can you also
> do a --trace on the 1.9 run (jruby --1.9 -S rake spec --trace).

Sure thing (the below is on the master of
<http://github.com/Chastell/art-decomp>):

shot@devielle:~/PhD/art-decomp$ jruby --1.9 -S rake spec --trace
(in /home/shot/PhD/art-decomp)
rake aborted!
no such file to load -- spec/rake/spectask
/home/shot/PhD/art-decomp/Rakefile:1:in `require'
/home/shot/PhD/art-decomp/Rakefile:1
/home/shot/PhD/art-decomp/Rakefile:2383:in `load'
/home/shot/.rvm/jruby-1.3.1/lib/ruby/gems/1.8/gems/rake-0.8.7/lib/rake.rb:2383:in `raw_load_rakefile'
/home/shot/.rvm/jruby-1.3.1/lib/ruby/gems/1.8/gems/rake-0.8.7/lib/rake.rb:2017:in `load_rakefile'
/home/shot/.rvm/jruby-1.3.1/lib/ruby/gems/1.8/gems/rake-0.8.7/lib/rake.rb:2068:in `standard_exception_handling'
/home/shot/.rvm/jruby-1.3.1/lib/ruby/gems/1.8/gems/rake-0.8.7/lib/rake.rb:2016:in `load_rakefile'
/home/shot/.rvm/jruby-1.3.1/lib/ruby/gems/1.8/gems/rake-0.8.7/lib/rake.rb:2000:in `run'
/home/shot/.rvm/jruby-1.3.1/lib/ruby/gems/1.8/gems/rake-0.8.7/lib/rake.rb:2068:in `standard_exception_handling'
/home/shot/.rvm/jruby-1.3.1/lib/ruby/gems/1.8/gems/rake-0.8.7/lib/rake.rb:1998:in `run'
/home/shot/.rvm/jruby-1.3.1/lib/ruby/gems/1.8/gems/rake-0.8.7/bin/rake:31
/home/shot/.rvm/jruby-1.3.1/lib/ruby/gems/1.8/gems/rake-0.8.7/bin/rake:19:in `load'
/home/shot/.rvm/jruby-1.3.1/bin/rake:19
shot@devielle:~/PhD/art-decomp$

Note that RSpec is there, under
/home/shot/.rvm/jruby-1.3.1/lib/ruby/gems/1.8/gems/rspec-1.2.9
What’s interesting, ‘jruby --1.8 -S gem list rspec’ says it’s
‘rspec (1.2.9)’, while ‘jruby --1.9 -S gem list rspec’ says it’s
‘rspec (local)’.

> Getting a reduced test case (like a minimal Rakefile with minimum
> specs showing the loading problem) would be super helpful also.

I think my Rakefile’s spec task is pretty minimal:
http://github.com/Chastell/art-decomp/blob/master/Rakefile :)

Are there any environmental variables that need
to be (un)set to rule out any interactions?

(Note that ‘jruby --1.8 -S rake spec’ does see the gems and works
properly-ish; it fails only because my Ruby-1.8-compatibility layer
is based on the backports gem, which relies on Process.exec.)

— Shot
--
‘“Cross-platform” used to mean “runs on Red Hat _and_
Slackware”. Now it means “runs on RHEL _and_ Fedora”.’
‘You are too kind. It means “worked in my system, morning and afternoon”.’
                    [Roger Burton West, Jarkko Hietaniemi, hates-software]

[Thomas E Enebo]

I can reproduce this.  I filed it as an issue:
http://jira.codehaus.org/browse/JRUBY-4156

I will try and look into this, but depending on whether I find
anything or whether it is too invasive it may not make it into 1.4
final (in fact expect that this will get fixed in 1.5 development --
master/trunk).

-Tom

On Sat, Oct 24, 2009 at 2:01 PM, Shot (Piotr Szotkowski) <shot@...> wrote:

> First: Thanks a lot for your reply! Apologies
> for the ambiguities, hopefully clarified below.
>
> Thomas E Enebo:
>
>> Hmmm, I think we will need to try and get some more information on
>> this one. Before we go any further trying to debug this, I noticed
>> one thing which looks wrong below. You have two lines like:
>
>> from /home/shot/.rvm/jruby-1.3.1/lib/ruby/site_ruby/1.8/rubygems/custom_require.rb:31:in `require'
>
>> It looks like you are loading at least some stuff from an older
>> version of jruby (1.3.1)? I have never used rvm so perhaps this
>> is ok, but it looks wrong at first glance.
>
> Argh, I’m sorry, I really should’ve clarified this: I’m using an old
> version of rvm, because never ones break for me. I simply deleted the
> old JRuby 1.3.1 files and put JRuby 1.4.0RC2 there. I also had this
> problem before 1.4.RC1 with a JRuby built from trunk.
>
>> Also if the rvm thing ends up being benign, then can you also
>> do a --trace on the 1.9 run (jruby --1.9 -S rake spec --trace).
>
> Sure thing (the below is on the master of
> <http://github.com/Chastell/art-decomp>):
>
> shot@devielle:~/PhD/art-decomp$ jruby --1.9 -S rake spec --trace
> (in /home/shot/PhD/art-decomp)
> rake aborted!
> no such file to load -- spec/rake/spectask
> /home/shot/PhD/art-decomp/Rakefile:1:in `require'
> /home/shot/PhD/art-decomp/Rakefile:1
> /home/shot/PhD/art-decomp/Rakefile:2383:in `load'
> /home/shot/.rvm/jruby-1.3.1/lib/ruby/gems/1.8/gems/rake-0.8.7/lib/rake.rb:2383:in `raw_load_rakefile'
> /home/shot/.rvm/jruby-1.3.1/lib/ruby/gems/1.8/gems/rake-0.8.7/lib/rake.rb:2017:in `load_rakefile'
> /home/shot/.rvm/jruby-1.3.1/lib/ruby/gems/1.8/gems/rake-0.8.7/lib/rake.rb:2068:in `standard_exception_handling'
> /home/shot/.rvm/jruby-1.3.1/lib/ruby/gems/1.8/gems/rake-0.8.7/lib/rake.rb:2016:in `load_rakefile'
> /home/shot/.rvm/jruby-1.3.1/lib/ruby/gems/1.8/gems/rake-0.8.7/lib/rake.rb:2000:in `run'
> /home/shot/.rvm/jruby-1.3.1/lib/ruby/gems/1.8/gems/rake-0.8.7/lib/rake.rb:2068:in `standard_exception_handling'
> /home/shot/.rvm/jruby-1.3.1/lib/ruby/gems/1.8/gems/rake-0.8.7/lib/rake.rb:1998:in `run'
> /home/shot/.rvm/jruby-1.3.1/lib/ruby/gems/1.8/gems/rake-0.8.7/bin/rake:31
> /home/shot/.rvm/jruby-1.3.1/lib/ruby/gems/1.8/gems/rake-0.8.7/bin/rake:19:in `load'
> /home/shot/.rvm/jruby-1.3.1/bin/rake:19
> shot@devielle:~/PhD/art-decomp$
>
> Note that RSpec is there, under
> /home/shot/.rvm/jruby-1.3.1/lib/ruby/gems/1.8/gems/rspec-1.2.9
> What’s interesting, ‘jruby --1.8 -S gem list rspec’ says it’s
> ‘rspec (1.2.9)’, while ‘jruby --1.9 -S gem list rspec’ says it’s
> ‘rspec (local)’.
>
>> Getting a reduced test case (like a minimal Rakefile with minimum
>> specs showing the loading problem) would be super helpful also.
>
> I think my Rakefile’s spec task is pretty minimal:
> http://github.com/Chastell/art-decomp/blob/master/Rakefile :)
>
> Are there any environmental variables that need
> to be (un)set to rule out any interactions?
>
> (Note that ‘jruby --1.8 -S rake spec’ does see the gems and works
> properly-ish; it fails only because my Ruby-1.8-compatibility layer
> is based on the backports gem, which relies on Process.exec.)
>
> — Shot
> --
> ‘“Cross-platform” used to mean “runs on Red Hat _and_
> Slackware”. Now it means “runs on RHEL _and_ Fedora”.’
> ‘You are too kind. It means “worked in my system, morning and afternoon”.’
>                    [Roger Burton West, Jarkko Hietaniemi, hates-software]
>
> -----BEGIN PGP SIGNATURE-----
> Version: GnuPG v1.4.9 (GNU/Linux)
>
> iEYEARECAAYFAkrjTx8ACgkQi/mCfdEo8UprYQCgqWzOwS2idJ24NcO6tGFMOwWc
> eE8An38NapjJxSOGlWV6jKHMV3/ke0SN
> =VVtB
> -----END PGP SIGNATURE-----
>
>

--
blog: http://blog.enebo.com       twitter: tom_enebo
mail: tom.enebo@...

---------------------------------------------------------------------
To unsubscribe from this list, please visit:

    http://xircles.codehaus.org/manage_email

[Shot (Piotr Szotkowski)]

Thomas E Enebo:

> I can reproduce this. I filed it as an issue:
> http://jira.codehaus.org/browse/JRUBY-4156

Thanks! I’ll added myself to the watchers.

> I will try and look into this, but depending on whether I find
> anything or whether it is too invasive it may not make it into
> 1.4 final (in fact expect that this will get fixed in 1.5 development
> -- master/trunk).

That’s ok, I’m happy with a trunk-built
JRuby. Thanks a lot for looking into this!

(Feel free to skip the below, I just need to vent this one time…)

// I have had a series of Ruby woes recently, with
// Set[2305860601668175887].hash == Set[2305860601668175887].hash
// being false in all MRI versions including current 1.9.1 svn code
// *and* in JRuby 1.3.1 when running Ruby 1.8 compatibility mode; these
// two hashes are equal only in JRuby 1.4 (which doesn’t work for me)
// and MRI 1.9.2 (both preview1 and current trunk), which, in turn,
// break with a lovely ‘no such file to load -- spec/spec_helper’ when
// I try to rake spec my code.

// As my PhD code heavily relies on Sets of Integers, I need to rewrite
// Set#hash – and I need it to be fast, so I wanted to go one step down
// and write it in Java, to see whether JVM is fast enough for my case.
// I guess I’m left with either going with C and Ruby InLine (I love the
// latter, but not really the former) or with D and RuDy (which I need
// to learn, but at least heard a lot of good things about). <sigh />

— Shot
--
A *huge* proportion of people cannot make *correct and accurate*
generalisations of principles. They have to learn everything as if
it’s an unrelated piece of crap, BECAUSE THEY ARE STUPID! PEOPLE ARE
STUPID! YES, THAT’S RIGHT, I’M SHOUTING NOW! AIEEEE!! PEOPLE ARE STUPID!
                                                         [Thorfinn, asr]

[Thomas E Enebo]

On Sat, Oct 24, 2009 at 2:39 PM, Shot (Piotr Szotkowski) <shot@...> wrote:
> Thomas E Enebo:
>
>> I can reproduce this. I filed it as an issue:
>> http://jira.codehaus.org/browse/JRUBY-4156
>
> Thanks! I’ll added myself to the watchers.

I added a hack to 1.4 which definitely fixes this issue for you.  It
is the wrong long term solution, but it only changes behavior for 1.9
and really loads gems exactly the same way 1.8 mode does so it is a
very very low risk change.   RC3 will likely be out tomorrow.  Give it
a try and report back.

Since this was a hack and it needs to be fixed the right way this is
broken on master and will stay that way until the right solution can
be found.

-Tom

>
>> I will try and look into this, but depending on whether I find
>> anything or whether it is too invasive it may not make it into
>> 1.4 final (in fact expect that this will get fixed in 1.5 development
>> -- master/trunk).
>
> That’s ok, I’m happy with a trunk-built
> JRuby. Thanks a lot for looking into this!
>
> (Feel free to skip the below, I just need to vent this one time…)
>
> // I have had a series of Ruby woes recently, with
> // Set[2305860601668175887].hash == Set[2305860601668175887].hash
> // being false in all MRI versions including current 1.9.1 svn code
> // *and* in JRuby 1.3.1 when running Ruby 1.8 compatibility mode; these
> // two hashes are equal only in JRuby 1.4 (which doesn’t work for me)
> // and MRI 1.9.2 (both preview1 and current trunk), which, in turn,
> // break with a lovely ‘no such file to load -- spec/spec_helper’ when
> // I try to rake spec my code.
>
> // As my PhD code heavily relies on Sets of Integers, I need to rewrite
> // Set#hash – and I need it to be fast, so I wanted to go one step down
> // and write it in Java, to see whether JVM is fast enough for my case.
> // I guess I’m left with either going with C and Ruby InLine (I love the
> // latter, but not really the former) or with D and RuDy (which I need
> // to learn, but at least heard a lot of good things about). <sigh />
>
> — Shot
> --
> A *huge* proportion of people cannot make *correct and accurate*
> generalisations of principles. They have to learn everything as if
> it’s an unrelated piece of crap, BECAUSE THEY ARE STUPID! PEOPLE ARE
> STUPID! YES, THAT’S RIGHT, I’M SHOUTING NOW! AIEEEE!! PEOPLE ARE STUPID!
>                                                         [Thorfinn, asr]
>
> -----BEGIN PGP SIGNATURE-----
> Version: GnuPG v1.4.9 (GNU/Linux)
>
> iEYEARECAAYFAkrjWAEACgkQi/mCfdEo8UoIhgCgvw9FMji/7vJRgkycOSGGUjJm
> XXIAn2nM4dDfCM8411rtAyLWn6d7d22j
> =Gw9n
> -----END PGP SIGNATURE-----
>
>

--
blog: http://blog.enebo.com       twitter: tom_enebo
mail: tom.enebo@...

---------------------------------------------------------------------
To unsubscribe from this list, please visit:

    http://xircles.codehaus.org/manage_email

[Shot (Piotr Szotkowski)]

Thomas E Enebo:

> On Sat, Oct 24, 2009 at 2:39 PM, Shot (Piotr Szotkowski) <shot@...> wrote:

>> Thomas E Enebo:

>>> I can reproduce this. I filed it as an issue:
>>> http://jira.codehaus.org/browse/JRUBY-4156

>> Thanks! I’ll added myself to the watchers.

> I added a hack to 1.4 which definitely fixes this issue for you. It is
> the wrong long term solution, but it only changes behavior for 1.9 and
> really loads gems exactly the same way 1.8 mode does so it is a very
> very low risk change. RC3 will likely be out tomorrow. Give it a try
> and report back.

I just did and it does indeed fixes gem loading for me – thanks a lot!

— Shot, off to make his code JRuby-compatible :)
--
The endless cycle of hope and disappointment has hardened cynicism.
I am afraid that we have reached the point where talking is useless.
                       [Jonathan Edwards, The Future of Programming]