New project: getting rid of IcedTea local patches

Martin Buchholz martinrb at google.com
Fri Apr 17 19:55:33 UTC 2009


On Thu, Apr 9, 2009 at 06:14, Andrew Haley <aph at redhat.com> wrote:
> Andrew Haley wrote:
>> We at Red Hat have noticed that the list of IcedTea local patches has
>> been getting large, with some local patches that should have been
>> pushed upstream.  Also, some IcedTea patches have been committed to
>> OpenJDK 7 but not OpenJDK 6.
>>
>> So, we're going to have a serious attempt to minimize the number of
>> IcedTea patches.  We'll create a Wiki page at icedtea.classpath.org
>> that is a list of the patches, and for every one we're going to
>> discuss whether it should stay IcedTea local, be pushed upstream to
>> OpenJDK, or be deleted altogether.  Then we will create OpenJDK
>> bugzilla entries for the patches that we think should be integrated
>> upstream and work with Sun engineers to get that done.
>
> The page is at http://icedtea.classpath.org/wiki/IcedTea_JDK6_Patches
> These are the patches in icedtea/patches and icedtea/patches/hotspot.

IcedTea folk and especially Andrew, thank you very much for this effort.

Here at Google, we also have some local patches and we have a similar
effort to push changes upstream.  It is probably a common experience
that patches are developed in a hurry to meet some release date,
and the upstream push effort is deferred to post-release.  This may result
in a delay before the rest of the openjdk community shares the patch,
but has the advantage that the patch is much more likely to be correct,
since it will already have had real users.

It occurs to me that there is a new reason to maintain a "buffer" forest
that contains patches on their way into Sun's master forest, namely
the periodic stabilization for jdk7 milestone releases.  It may make sense
for Google (i.e. me) to push changes
into the icedtea upstream-directed forest
if openjdk forests are in quiescent mode.

Martin



More information about the build-dev mailing list