Producing community binaries for OpenJDK

[Martijn Verburg]

Hi Jiri,

Thanks - we'll be looking to utilise Cloudbee's service in the initial
instance but if we can't do that then yes we'll be looking to host a
small cluster of hardware to run the builds.
Cheers,
Martijn


On 14 March 2017 at 12:11, Jiri Vanek <jvanek at redhat.com> wrote:
> Hi!
>
> As I consider this feature really missing, I would like to dontate my $0.02:
>
> If you - Oracle, community, IBM whoever rich enough -  will provide..well
> donate -  an reasonable fast machine[0], I will offer my time and my code[1]
> to establish proper pull-snapshot-build-offerImages (later also test) for
> all reasonable forests on http://hg.openjdk.java.net and
> http://icedtea.classpath.org/hg/.
> If this is working fine, it would be awesome to include it like some
> x.openjdk.java.net service.
> I do this already for years, to have pre-releases for fedora packages.
>
> J.
>
>
> [0]usually something like x3500 tower server with 2x2.5gh 8core xeon and
> about 72Gb ram, which can run 6 builders in time -  is enough.
>
> [1] 99% is done by this jenkins plugin -
> https://github.com/judovana/jenkins-scm-koji-plugin. the missing 1% is
> jenkins job configuration, which does not much sense to push public  without
> comunity around. it is forest-pull script and absolute minimal build scripts
> to provide both fastdebug and release static builds for linuxes and windows.
> Some 20kb with a patch allowing generic builds on "oldest possible linux"