Buildr support

Hi guys,

we are trying to figure a way to get Buildr out of the door.

First, it looks like we have failures with JRuby 1.4.1 on Windows 7:
https://issues.apache.org/jira/browse/BUILDR-443
=> Could you take a look at the file attached and see if those errors
are
fixed in 1.5 ?

Second, we cannot run with JRuby 1.5 (this happens on Mac or Windows 7):
https://issues.apache.org/jira/browse/BUILDR-441
which points to:
https://issues.apache.org/jira/browse/BUILDR-418
=> Anybody with FFI knowledge I can chat with on that ?
https://issues.apache.org/jira/browse/BUILDR-442
=> Did you guys make changes to Rake, by monkey-patching it or enhancing
it
? This error is weird.

We can probably concentrate on supporting JRuby 1.5 on Windows 7 only.

Any help or comment are very welcome.

I am toulmean on #jruby, feel free to ping me there to chat about this.

Thanks,

Antoine

On 21 May 2010 11:26, Antoine T. [email protected] wrote:

=> Anybody with FFI knowledge I can chat with on that ?
That patch looks to be ok. JRuby 1.5 (and ruby-ffi 0.6) changed to
require you to specify a library, instead of just using the current
process by default. (the old way was causing people to rely on the
libs loaded into the ruby vm, and they weren’t always there on jruby,
so it was changed to force people to explicitly state the lib they
expected to get functions from).


To unsubscribe from this list, please visit:

http://xircles.codehaus.org/manage_email

Hello!

On Thu, May 20, 2010 at 8:26 PM, Antoine T.
[email protected] wrote:

Hi guys,
we are trying to figure a way to get Buildr out of the door.
First, it looks like we have failures with JRuby 1.4.1 on Windows 7:
https://issues.apache.org/jira/browse/BUILDR-443
=> Could you take a look at the file attached and see if those errors are
fixed in 1.5 ?

There’s about 43 of them, yeah? Have they ever passed? I didn’t scroll
through the whole list of failures (it was very long).

Second, we cannot run with JRuby 1.5 (this happens on Mac or Windows 7):
https://issues.apache.org/jira/browse/BUILDR-441
which points to:
https://issues.apache.org/jira/browse/BUILDR-418
=> Anybody with FFI knowledge I can chat with on that ?

Wayne covered this. However if I set it to always explicitly load C
library, I get the following now, perhaps with newer RubyGems (not
sure if this affects RubyGems 1.3.6 or just 1.3.7 which we now have on
master:

~/projects/jruby âž” jruby -rubygems -e “p Gem::Requirement::OP_RE”
-e:1: uninitialized constant Gem::Requirement::OP_RE (NameError)

That shows up here in Buildr:

~/projects/jruby/minitwitter âž” buildr
/Users/headius/projects/jruby/lib/ruby/gems/1.8/gems/rake-0.8.7/lib/rake.rb:2503:in
`const_missing’: uninitialized constant Gem::Requirement::OP_RE
(NameError)
from
/Users/headius/projects/jruby/lib/ruby/gems/1.8/gems/buildr-1.3.5-java/lib/buildr/packaging/version_requirement.rb:24

Not sure what you’re using that for or if it affects 1.5.

https://issues.apache.org/jira/browse/BUILDR-442
=> Did you guys make changes to Rake, by monkey-patching it or enhancing it
? This error is weird.

I don’t believe so, and in fact we install rake from a stock gem
during our build. The specific Ant support isn’t loaded unless you
require ‘ant’…maybe we have a library name conflict?

  • Charlie

To unsubscribe from this list, please visit:

http://xircles.codehaus.org/manage_email

On Thu, May 20, 2010 at 9:22 PM, Charles Oliver N.
[email protected] wrote:

https://issues.apache.org/jira/browse/BUILDR-442
=> Did you guys make changes to Rake, by monkey-patching it or enhancing it
? This error is weird.

I don’t believe so, and in fact we install rake from a stock gem
during our build. The specific Ant support isn’t loaded unless you
require ‘ant’…maybe we have a library name conflict?

Reproduced here with 1.6.0.dev of JRuby. I’ll give it a shot.

  • Charlie

To unsubscribe from this list, please visit:

http://xircles.codehaus.org/manage_email

Looks like a regression in JRuby 1.5 in Array#find:
http://jira.codehaus.org/browse/JRUBY-4805

It’s part of the methods normally inherited from Enumerable that we
fast-pathed. The fast-path for at least “find” isn’t safe in the
presence of mutation during iteration. Others may have similar bugs.

Mutation during these methods probably falls into a grey area, but
it’s definitely a behavioral regression and it doesn’t match MRI.

On Thu, May 20, 2010 at 9:32 PM, Charles Oliver N.
[email protected] wrote:

Reproduced here with 1.6.0.dev of JRuby. I’ll give it a shot.

  • Charlie

To unsubscribe from this list, please visit:

http://xircles.codehaus.org/manage_email