You are viewing a plain text version of this content. The canonical link for it is here.
Posted to users@buildr.apache.org by Niklaus Giger <ni...@member.fsf.org> on 2012/05/01 22:07:28 UTC
Enable building jekyll and rdoc under more rubies
Hi
I got frustrated that usinge jruby there were a lot of rake target not working
in my buildr checkout.
Therefore I invested some time to fix the problem. Please look at my patch
under https://github.com/apache/buildr/pull/5
Also I did set up some CI builds on my jenkins, where you can see the results
without my patch under http://ngiger.dyndns.org/jenkins/job/buildr-matrix/
and from my fork under http://ngiger.dyndns.org/jenkins/job/buildr-matrix-
ngiger/
The RSpec results are not too bad. Usually just 3 or 4 failures of over 2200
examples, e.g. http://ngiger.dyndns.org/jenkins/job/buildr-matrix-
ngiger/Rake_Target=spec,rubies=ruby-1.9.2-p320/ws/_reports/specs.html
The Jenkins build for buildr.apache.org at https://builds.apache.org/ seem a
bit outdated, and I think a matrix based build as in my example is quite nice
to read and could be an inspiration to update the buildr at
https://builds.apache.org/
My Matrix build misses tests on different platforms, mainly because I don't
have enough CPUs/RAM to run the tests also under Windows and MacOSX.
Regards
Niklaus
Re: Enable building jekyll and rdoc under more rubies
Posted by Antoine Toulme <an...@lunar-ocean.com>.
Alex Boisvert is the PMC of Buildr.
Yes, you need committer rights to access the build machines - so working on those means you need to be voted a committer first.
They are not easy to work with or maintain mainly because they were not much used for ruby builds in the path.
RVM changed that partly, but it's still hard to maintain the different combinations.
Thanks,
Antoine
On Thursday, May 3, 2012 at 4:49 AM, Niklaus Giger wrote:
> Hi Antoine
>
> Opened issue at https://issues.apache.org/jira/browse/BUILDR-640, added patch
> and assigned copyright to the ASF.
>
> I agree that sometimes it is hard work to maintain a Jenkins CI. Usually the
> benefits outweighted the inconvenience, if the project members see the
> benefits and are willing to accept patches/suggestion to make the Jenkins run.
>
> I had a look at "Jenkins wiki page" under https://builds.apache.org/ and found
> the policy very reasonable. (Eg. one can build under Ubuntu, Windows, MacOSX
> and Solaris.) But it states "PMC chairs can grant access to Jenkins to any
> committer". I am neither a commiter nor do I know who are the PMC for
> buildr.apache.org (http://buildr.apache.org).
>
> If you (or somebody) else gives me the necessary rights, I am willing to do my
> best to maintain a Jenkins CI for buildr for at least one year. (Hoping that
> it will be useful for a bright future and more frequent releases of buildr.)
>
> In the mean time I will continue to run my Jenkins CI on my home server,
> checking once a day for modifications.
>
> Best regards
>
> Niklaus
>
> > Hi Niklaus,
> >
> > please see my comments inline:
> >
> > On Tuesday, May 1, 2012 at 1:07 PM, Niklaus Giger wrote:
> > > Hi
> > >
> > > I got frustrated that usinge jruby there were a lot of rake target not
> > > working in my buildr checkout.
> > >
> > > Therefore I invested some time to fix the problem. Please look at my
> > > patch under https://github.com/apache/buildr/pull/5
> > >
> >
> >
> > I haven't tried it but it looks good.
> > We can't accept pull requests but if you attach it as a patch in jira and
> > check that you give copyright to the ASF it should be ok.
> >
> > > Also I did set up some CI builds on my jenkins, where you can see the
> > > results without my patch under
> > > http://ngiger.dyndns.org/jenkins/job/buildr-matrix/ and from my fork
> > > under http://ngiger.dyndns.org/jenkins/job/buildr-matrix- ngiger/
> > >
> > > The RSpec results are not too bad. Usually just 3 or 4 failures of over
> > > 2200 examples, e.g. http://ngiger.dyndns.org/jenkins/job/buildr-matrix-
> > > ngiger/Rake_Target=spec,rubies=ruby-1.9.2-p320/ws/_reports/specs.html
> > >
> >
> >
> > That's not good at all :) All specs must pass for a release to be made so
> > it usually gets done at some point. We are lagging behind there.
> >
> > > The Jenkins build for buildr.apache.org at https://builds.apache.org/
> > > seem a bit outdated, and I think a matrix based build as in my example
> > > is quite nice to read and could be an inspiration to update the buildr
> > > at
> > > https://builds.apache.org/
> > >
> >
> >
> > It's completely outdated. The jenkins server is very hard to maintain and
> > we historically made it work, though it became very quickly a hassle. We
> > need to move out from there and find a better strategy. We don't have
> > resources to host a jenkins instance at this point. If you feel like
> > taking this part over, that would be most awesome.
> >
> > > My Matrix build misses tests on different platforms, mainly because I
> > > don't have enough CPUs/RAM to run the tests also under Windows and
> > > MacOSX.
> > >
> >
> >
> > It's better than what we have.
> >
> > > Regards
> > >
> > > Niklaus
Re: Enable building jekyll and rdoc under more rubies
Posted by Niklaus Giger <ni...@member.fsf.org>.
Hi Antoine
Opened issue at https://issues.apache.org/jira/browse/BUILDR-640, added patch
and assigned copyright to the ASF.
I agree that sometimes it is hard work to maintain a Jenkins CI. Usually the
benefits outweighted the inconvenience, if the project members see the
benefits and are willing to accept patches/suggestion to make the Jenkins run.
I had a look at "Jenkins wiki page" under https://builds.apache.org/ and found
the policy very reasonable. (Eg. one can build under Ubuntu, Windows, MacOSX
and Solaris.) But it states "PMC chairs can grant access to Jenkins to any
committer". I am neither a commiter nor do I know who are the PMC for
buildr.apache.org.
If you (or somebody) else gives me the necessary rights, I am willing to do my
best to maintain a Jenkins CI for buildr for at least one year. (Hoping that
it will be useful for a bright future and more frequent releases of buildr.)
In the mean time I will continue to run my Jenkins CI on my home server,
checking once a day for modifications.
Best regards
Niklaus
> Hi Niklaus,
>
> please see my comments inline:
>
> On Tuesday, May 1, 2012 at 1:07 PM, Niklaus Giger wrote:
> > Hi
> >
> > I got frustrated that usinge jruby there were a lot of rake target not
> > working in my buildr checkout.
> >
> > Therefore I invested some time to fix the problem. Please look at my
> > patch under https://github.com/apache/buildr/pull/5
>
> I haven't tried it but it looks good.
> We can't accept pull requests but if you attach it as a patch in jira and
> check that you give copyright to the ASF it should be ok.
>
> > Also I did set up some CI builds on my jenkins, where you can see the
> > results without my patch under
> > http://ngiger.dyndns.org/jenkins/job/buildr-matrix/ and from my fork
> > under http://ngiger.dyndns.org/jenkins/job/buildr-matrix- ngiger/
> >
> > The RSpec results are not too bad. Usually just 3 or 4 failures of over
> > 2200 examples, e.g. http://ngiger.dyndns.org/jenkins/job/buildr-matrix-
> > ngiger/Rake_Target=spec,rubies=ruby-1.9.2-p320/ws/_reports/specs.html
>
> That's not good at all :) All specs must pass for a release to be made so
> it usually gets done at some point. We are lagging behind there.
>
> > The Jenkins build for buildr.apache.org at https://builds.apache.org/
> > seem a bit outdated, and I think a matrix based build as in my example
> > is quite nice to read and could be an inspiration to update the buildr
> > at
> > https://builds.apache.org/
>
> It's completely outdated. The jenkins server is very hard to maintain and
> we historically made it work, though it became very quickly a hassle. We
> need to move out from there and find a better strategy. We don't have
> resources to host a jenkins instance at this point. If you feel like
> taking this part over, that would be most awesome.
>
> > My Matrix build misses tests on different platforms, mainly because I
> > don't have enough CPUs/RAM to run the tests also under Windows and
> > MacOSX.
>
> It's better than what we have.
>
> > Regards
> >
> > Niklaus
Re: Enable building jekyll and rdoc under more rubies
Posted by Antoine Toulme <an...@lunar-ocean.com>.
Hi Niklaus,
please see my comments inline:
On Tuesday, May 1, 2012 at 1:07 PM, Niklaus Giger wrote:
> Hi
>
> I got frustrated that usinge jruby there were a lot of rake target not working
> in my buildr checkout.
>
> Therefore I invested some time to fix the problem. Please look at my patch
> under https://github.com/apache/buildr/pull/5
>
>
I haven't tried it but it looks good.
We can't accept pull requests but if you attach it as a patch in jira and check that you give copyright to the ASF it should be ok.
>
> Also I did set up some CI builds on my jenkins, where you can see the results
> without my patch under http://ngiger.dyndns.org/jenkins/job/buildr-matrix/
> and from my fork under http://ngiger.dyndns.org/jenkins/job/buildr-matrix-
> ngiger/
>
> The RSpec results are not too bad. Usually just 3 or 4 failures of over 2200
> examples, e.g. http://ngiger.dyndns.org/jenkins/job/buildr-matrix-
> ngiger/Rake_Target=spec,rubies=ruby-1.9.2-p320/ws/_reports/specs.html
>
>
That's not good at all :) All specs must pass for a release to be made so it usually gets done at some point. We are lagging behind there.
>
> The Jenkins build for buildr.apache.org at https://builds.apache.org/ seem a
> bit outdated, and I think a matrix based build as in my example is quite nice
> to read and could be an inspiration to update the buildr at
> https://builds.apache.org/
>
>
It's completely outdated. The jenkins server is very hard to maintain and we historically made it work, though it became very quickly a hassle. We need to move out from there and find a better strategy.
We don't have resources to host a jenkins instance at this point. If you feel like taking this part over, that would be most awesome.
>
> My Matrix build misses tests on different platforms, mainly because I don't
> have enough CPUs/RAM to run the tests also under Windows and MacOSX.
>
>
It's better than what we have.
>
>
> Regards
>
> Niklaus