|
Notes |
|
|
(0009584)
|
|
bwalton
|
|
2012-02-07 01:52
|
|
I had a working package set for 1.9 a long time back but couldn't release it due to certain blockers. The last time I updated the 1.9 recipe, the test suite was failing fairly badly.
I'll update again and see what happens...if required, I'll try on sol10 only, dropping support for 9.
I'd like to have this out there too but it hasn't been a priority for me lately. Thanks for the push! |
|
|
|
(0009585)
|
|
jcraig
|
|
2012-02-07 01:59
|
|
Pushes are easy! :)
I have some developers that are starting a skunk works project and want to use ruby 1.9 to do it. It will be used in a Solaris 11 x86 zone. The first version is running with the 1.8 from OpenCSW but they would like to use some newer features of 1.9. Let me know how I can help.
Jon |
|
|
|
(0009591)
|
|
bwalton
|
|
2012-02-12 02:37
|
|
Status update:
The recipe was using gcc4 and on i386, things are in good shape. On sparc, the binaries were generating bus errors. I've switched back to sun CC and sparc looks better (using -xmemalign=li). I'll see what things are like on i386 with sun CC.
The errors on i386 are partially ignorable...once we have a released 1.9, the webrick.cgi tests should work. They're started with a cleaned environment and therefore can't leverage LD_PRELOAD set by the test suite.
The only error that seems to be of real significance is a gdbm error...
I think we can get test packages put together in fairly short order. |
|
|
|
(0009593)
|
|
bwalton
|
|
2012-02-12 14:35
|
|
One troubling issue that affects both platforms with sun pro:
29) Failure:
test_singleton_method(TestGc) [/home/bwalton/opencsw/ruby19/trunk/work/solaris10
-i386/build-isa-pentium_pro/ruby-1.9.3-p0/test/ruby/test_gc.rb:71]:
[ruby-dev:42832].
Exception raised:
<#<RuntimeError: obj.foo is called, but this is obj.bar>>. |
|
|
|
(0009595)
|
|
bwalton
|
|
2012-02-12 18:24
|
|
|
Ok, got a patch for the GC test failure...cherry picked from upstream. |
|
|
|
(0009663)
|
|
bwalton
|
|
2012-02-26 14:30
|
|
|
I'm pushing the first ruby191 packages to unstable now. I still need to add alternatives support and the generic ruby19 meta-package, but these should give your team something to work with. Feedback is definitely welcome. If you (or your team) would like to see some structural changes to the package, please let me know. |
|
|
|
(0009667)
|
|
bwalton
|
|
2012-02-27 04:37
|
|
I just pushed CSWruby19{,-ri,-dev} that are meta packages to pull in the 191 versions of the same. In the case of CSWruby19 and CSWruby19-ri, they also deliver files in bin/ to allow for using /opt/csw/bin/ruby19 as the hashbang, etc.
With these packages, getting ruby19 on a box should be doable with: pkgutil -i CSWruby19.
The alternatives support lives in CSWruby191. You can make ruby 1.9 the default by toggling the alternative if you want to.
Feedback welcome. |
|
|
|
(0009673)
|
|
jcraig
|
|
2012-02-27 20:42
|
|
The ruby191 package date is 2/26 and appears to predate your alternatives work on 2/27 as I can find no /opt/csw/share/alternatives files.
Do you plan to split the gem stuff out or leave it part of the ruby19 package. I was building some native gems using the gcc version of the ruby package just to avoid installing spro.
Thanks for all the work you've put into this. I do appreciate it. |
|
|
|
(0009674)
|
|
bwalton
|
|
2012-02-28 01:32
|
|
Ooops. I didn't push the full update. I'm pushing a matching set net, so the alternatives will be available shortly.
I'd planned to leave it as part of the primary package but I still need to add the gcc4 version of rbconfig.rb and setup the alternatives for that. I guess that's a priority item on your end? I can try to get that out quickly but I wanted to focus on a working package before getting 'fancy!' :)
Thanks
-Ben |
|
|
|
(0009829)
|
|
jcraig
|
|
2012-04-24 05:38
|
|
So far ruby is working well. One thing missing from the alternatives is the gem191 link to gem.
Thanks,
Jon |
|