If #2429 (showstopper) is not fixed prior to the 9k preview release, Windows should not be part of the preview release and this should be in the release notes.

@cmichon At the moment we do want Windows to work for the release, but we're limited in what we can do to get there. Only @enebo has a Windows dev machine to use, though I have some Windows licenses on the way to remedy that.

Thanks for pointing it out.

I can provide feedbacks on regular basis with a custom test suite.

How can I find all recent daily snapshots? I would need all of them, not
just the latest CI build.
On 12 Jan, 2015 8:56 PM, "Charles Oliver Nutter" notifications@github.com
wrote:

@cmichon https://github.com/cmichon At the moment we do want Windows
to work for the release, but we're limited in what we can do to get there.
Only @enebo https://github.com/enebo has a Windows dev machine to use,
though I have some Windows licenses on the way to remedy that.

Thanks for pointing it out.

—
Reply to this email directly or view it on GitHub
#2445 (comment).

Commits leading up to a2ef35f have gotten Windows working well enough to install rails, generate and scaffold an app and serve it successfully. I'm sure there will be other issues, but that's a lot working.

yes today's CI snapshot confirms the original issue is gone.

Yet, installing the gem awesome_print gave this interesting output:

jgem ins awesome_print
Fetching: awesome_print-1.6.1.gem (100%)
Successfully installed awesome_print-1.6.1
ERROR: While executing gem ... (Errno::ENOENT)
No such file or directory - stty

grep -lr stty . (inside jruby-9.0.0.0-SNAPSHOT)
./bin/jruby
./bin/jruby.bash
./lib/ruby/stdlib/io/console.rb
./lib/ruby/stdlib/rake/application.rb
./lib/ruby/stdlib/rubygems/user_interaction.rb

Bound to happen: no stty in Windows...

Looking at io/console.rb, it seems the code fails on Windows and it behaves
as if stty is present on the system... I'll look more into it tomorrow.

On Wed, Jan 14, 2015 at 8:23 PM, Charles Oliver Nutter <
notifications@github.com> wrote:

Commits leading up to a2ef35f
a2ef35f
have gotten Windows working well enough to install rails, generate and
scaffold an app and serve it successfully. I'm sure there will be other
issues, but that's a lot working.

—
Reply to this email directly or view it on GitHub
#2445 (comment).

Christian

@cmichon Huh yeah, that should work. I took a bare JRuby clone all the way to a running Rails app, which involved a lot of RubyGems (which loads io/console). Let me know what you see and file a bug if you can narrow something down.

@headius you have cygwin or mingw commands in your path? The code path for io/console.rb looks like windows will follow the else and try to execute stty -a

@enebo It's possible, but it would have had to come in with the git command. I started from a bare Windows 2012 server. I don't think so, because even simple things like ls didn't work but git did.

At @enebo's suggestion, I audited core class method definitions in JRuby and MRI. I added a bunch of small ones that were missing, deleted some, moved some, and changed visibility on some. The remaining (coarse-grained) items are here:

• Enumerator#feed, #next_values, #prev_values missing
• ObjectSpace::WeakMap#each and Enumerable inclusion
• ObjectSpace::count_objects unimplemented
• Process::daemon unimplemented
• Process::exec missing (is this just Kernel#exec?)
• Process::getsid unimplemented
• Process::setproctitle
• RubyVM namespace
• Thread#handle_interrupt is not yet fully functional

@Who828 and I added a bunch of missing methods and removed some others in commits today.

Looking pretty good :-)

Oops, @Who828 added Array#any? and Hash#any? so I'll update my comment above.

FYI, current plan is pre1 release on Monday.

@chrisseaton I heard you're traveling today, but please comment on the bug with any details about JRuby+Truffle in pre1.

I've started a wiki page for the release notes: https://github.com/jruby/jruby/wiki/JRuby-9.0.0.0.pre1-Release-Notes

I've added a few sentences on Truffle. It basically says normal users shouldn't even attempt to run their applications on it.

Truffle is blocked on #2466, but @mkristian is already looking at it. We'll freeze Truffle development and start double checking to make sure it's ready for release on Monday.

Truffle is ready to go, apart from @mkristian thinks we need to put Truffle into Maven Central as all JRuby dependencies are supposed to be available there (#2474). I can look into that on Monday when the relevant people are around.

@headius I think you meant peek_values, not prev_values for Enumerator. (prev_values doesn't exists in MRI).

@Who828 You're right, I'll fix that.

It is out!