1.5.6 final release

[benlangfeld]

master contains a great many bug fixes, especially with regards to JRuby compatibility, which are desperately needed. 1.5.5 was released more than 5 months ago. I think it's time for a bugfix release (1.5.6) and don't see any new bugs on master, or anything else that should present a problem.

Can we get a new release?

[johntdyer]

+1

[soulcutter]

👍

[emcgee]

+1

[vindir]

+1

[lephyrius]

+1

[lpradovera]

+1 and then some.

[bklang]

We have a widely deployed framework which depends on Nokogiri and people want to run it on JRuby. There are several bugfixes in Nokogiri that are blocking that with our framework currently, but are fixed in the unreleased software. Right now we have to tell these people to use the Nokogiri prerelease, which obviously leaves people feeling less than comfortable. Are there any remaining issues ahead of releasing this bugfix version? If we can help, please let us know so everyone can benefit from the improvements already slated to be in 1.5.6.

[benlangfeld]

@flavorjones, @yokolet: Ping

[pixe1f10w]

+1

[benlangfeld]

@flavorjones Anything in particular blocking this? Anything I can help with to prep for a release? This is increasingly becoming a problem for us and really seems like it should be a trivial pushing of the release button.

[lpradovera]

I would gladly help too.

[benlangfeld]

I'm unsure what the problem is here. If it's just a lack of time to actually do the release, that's one thing, but I'd at least like confirmation that a release will come soon.

This is really hurting us and it feels like I've made a big effort to help improve some of Nokogiri's behaviour on JRuby, but seeing the fruits of that effort is being blocked by something as trivial as a patch release.

[flavorjones]

Lack of time is preventing me from doing the full regression test suite on all rubies, valgrinding, etc.

What's preventing everyone from releasing with the RC currently up on rubygems? I don't understand the pain here.

[benlangfeld]

I can't tag my gem to depend on your RC. That would be all sorts of screwed up.

[benlangfeld]

Out of interest, where would I find that "full regression test suite"? Is that not what TravisCI is running? Can we not get those things automated?

[flavorjones]

Ben,

Before I sign off on a full release, I run valgrind against it to ensure no memory errors or leaks have been introduced since the last version. Travis does not support valgrind at this time.

Why is it "screwed up" to rely on an RC of Nokogiri, if you're asking for that same commit to be released as a full version? I don't understand why there's pain here.

[benlangfeld]

It's designated as an RC. That means it has yet to be signed off for full release, which suggests there could be something wrong with it that I am not aware of. It would be irresponsible of me to force users of my code to use a version of Nokogiri which has not yet been signed off. That's precisely why I've been asking what is preventing a release; it really is not clear.

[flavorjones]

1.5.6 has been released.

[benlangfeld]

Thanks Mike.

libgit2 appears to run Valgrind on Travis: libgit2/libgit2#816. I'll see if I can get this working with Nokogiri to smooth out this process.

[flavorjones]

We use rake test:valgrind under the hood (the hoe-debugging gem) to test. The log files need to be captured for inspection right now.