Tomas.Matousek at microsoft.com
Mon Oct 31 20:10:22 EDT 2011
There is a lot of work to do around exceptions in general. I wouldn't start digging into it if I were you :). I'm thinking of some core changes in how we map Ruby to .NET exceptions... it will take me some time to get to it though. I would not worry about all these Errno exceptions for now unless your application is blocked by some specific bug. Certainly not if a few specs fail here and there (we have thousands of specs failing, which might be more important to fix).
Time to file a bug on CRuby to get some input from MRI guys what they think the behavior should really be if it doesn't make sense. Or RubySpec guys if the MRI behavior is reasonable.
From: ironruby-core-bounces at rubyforge.org [mailto:ironruby-core-bounces at rubyforge.org] On Behalf Of Orion Edwards
Sent: Monday, October 31, 2011 4:40 PM
To: ironruby-core at rubyforge.org
Subject: Re: [Ironruby-core] RubySpecs
Thanks for that.
After the code review comments about Exception#== I was running the Exception specs. 3 fail against ruby 1.9.2p290 (2011-07-09) [i386-mingw32]
I don't have access to a linux/mac system at work to check if they pass on non-windows platforms, but it doesn't look like there's anything platform related in those errors, so I'd doubt it.
It appears that there's a lot more work to be done in IronRuby particularly relating to Errno and SystemCallErrors - I could do this work, but there's a blocking problem:
- Various bits of IronRuby core code call helper methods such as RubyExceptions.CreateEINVAL, RubyExceptions.CreateEEXIST, etc, etc. These are supposed to return Errno::EINVAL, but they can't, because the Errno classes are defined in IronRuby.Libraries and the RubyExceptions.CreateXYZ methods are defined in IronRuby.dll. We'd have to move the Errno stuff out of Libraries and into IronRuby.dll I think
From: Tomas Matousek <Tomas.Matousek at microsoft.com<mailto:Tomas.Matousek at microsoft.com>>
To: "ironruby-core at rubyforge.org<mailto:ironruby-core at rubyforge.org>" <ironruby-core at rubyforge.org<mailto:ironruby-core at rubyforge.org>>
Date: 01/11/2011 10:16 a.m.
Subject: Re: [Ironruby-core] RubySpecs
Sent by: ironruby-core-bounces at rubyforge.org<mailto:ironruby-core-bounces at rubyforge.org>
1) Partly. We are in progress of updating them. Since we've added more specs and sometimes corrected existing ones this needs to be done with care. The current progress is captured here:
After we are done with this process of bringing the specs up to date we can submit a path to RubySpec containing the changed we made. The plan is also to make the RubySpec dir a submodule so that merging with RubySpec is easier.
2) Or maybe we should just remove MRI completely? You can always use one on your system. I need to check if we need CRuby for anything in the infrastructure.
3) Well, that would be a question for RubySpec maintainers. I found some specs be failures Windows specific, which means the specs are not properly written. Feel free to contribute to RubySpecs git repo - I bet they are happy to accept fixes.
4) Well, you need to figure out if the behavior makes sense. If it doesn't feel free to file a bug on CRuby and let them decide if the behavior is intentional or not. Then CRuby might get fixed, specs might get fixed, or IronRuby might get fixed.
Which specs particularly are you working on? Is it marshal related?
From: ironruby-core-bounces at rubyforge.org<mailto:ironruby-core-bounces at rubyforge.org> [mailto:ironruby-core-bounces at rubyforge.org] On Behalf Of Orion Edwards
Sent: Monday, October 31, 2011 1:19 PM
To: ironruby-core at rubyforge.org<mailto:ironruby-core at rubyforge.org>
Subject: [Ironruby-core] RubySpecs
I've just run mspec under the Languages\Ruby\Tests\mspec\rubyspec\core folder against MRI 1.9.2p0 (which is checked into the ironlanguages-main git repo), and it crashes with a segfault in putc_spec.rb
I then tried running against MRI 1.9.2p290 and I get the following:
1518 files, 9618 examples, 28229 expectations, 339 failures, 256 errors
Running against the dev build of IronRuby stalls running tests in the array subfolder, so I can't even complete the run.
I'd like to try fixing some of the spec failures that IronRuby has, but these results raise several questions:
1. Are the specs in the mspec\rubyspec folder up to date? How would someone find this out and/or update them?
- Also, I added a few specs the other day for the marshalling code. Do these need to somehow get pushed upstream into some "master" rubyspec repository??
2. Shouldn't we check the latest build of MRI into the ironlangauges git-repo? Either the latest 1.9.2 or the just-released 1.9.3 instead?
3. Isn't MRI the "definitive" ruby? How can the specs be failing against MRI?
4. If I update some IronRuby code to pass the specs, how do I know that the specs are even correct?
Ironruby-core mailing list
Ironruby-core at rubyforge.org<mailto:Ironruby-core at rubyforge.org>
-------------- next part --------------
An HTML attachment was scrubbed...
More information about the Ironruby-core