hadoop-common-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Andrew Wang <andrew.w...@cloudera.com>
Subject Re: Setting JIRA fix versions for 3.0.0 releases
Date Fri, 26 Aug 2016 00:06:53 GMT
I've taken a look at the Common and HDFS compat issues.

* A bunch of them are from the known removals of the deprecated Metrics 1
and RCC. JACC understands and can filter @Deprecated annotations, but
HADOOP-7266 isn't in 2.7.2, and apparently JACC doesn't look at the class
annotation when telling you some methods were removed. Filed:
https://github.com/lvc/japi-compliance-checker/issues/33
* The tool doesn't detect when you add a type parameter to a
super-interface, which I think is compatible. Filed a JACC issue:
https://github.com/lvc/japi-compliance-checker/issues/32
* It did find HDFS-10793, a legit audit logger issue which is an issue for
2.8.0 also. Manoj is closing on this, we're just waiting on a precommit run
before committing.

This list also includes LimitedPrivate classes, so things like DU and
KerberosName show up here too. I think it's good to check LimitedPrivate
too since practically speaking many of these should be Public, but I can
generate a Public only report if desired.

If someone wants to take a look at the YARN changes, that'd be swell. Since
things overall look good though, I'm going get back to rebranching for
3.0.0-alpha1 and the corresponding fix version updates.

On Thu, Aug 25, 2016 at 11:44 AM, Andrew Wang <andrew.wang@cloudera.com>
wrote:

> Sean gave me some pointers on using Java ACC, I've made a report using the
> script I've been working on at YETUS-445:
>
> http://home.apache.org/~wang/h3/report.html
>
> Invocation was something like this:
>
> -> % ~/dev/yetus/check-java-compatibility/checkjavacompatibility.py
> --annotation org.apache.hadoop.classification.InterfaceAudience.Public
> --annotation org.apache.hadoop.classification.InterfaceAudience.LimitedPrivate
> --include "hadoop.*" branch-2.7.2 trunk
>
> I think this report is easier to consume than JDiff. If we like it, I can
> generate one for 2.7.2 -> 2.8.0, and it should be easy to automate via
> Jenkins as well.
>
>
> On Tue, Jul 26, 2016 at 11:28 AM, Wangda Tan <wheeleast@gmail.com> wrote:
>
>> Hi all,
>>
>> I'm trying to generate JDiff for sub projects of Hadoop, some updates:
>> *- Common*: JDiff cannot be generated , filed https://issues.apache.or
>> g/jira/browse/HADOOP-13428 and debugging that.
>> - *HDFS*: It pointed to a older version (2.6.0), we need to upgrade it
>> to the latest stable release (probably 2.7.3, it is mostly done). filed
>> https://issues.apache.org/jira/browse/HDFS-10692 and will work on that
>> - *YARN*: JDiff can be generated, we need to do analysis for it.
>> - *MR*: JDiff cannot be generated, https://issues.apac
>> he.org/jira/browse/MAPREDUCE-6310 is tracking it.
>>
>> I'm currently actively working on these items, any helps will be
>> appreciated.
>>
>> Thanks,
>> Wangda
>>
>> On Tue, Jul 26, 2016 at 11:16 AM, Vinod Kumar Vavilapalli <
>> vinodkv@apache.org> wrote:
>>
>>> +1
>>>
>>> Thanks
>>> +Vinod
>>>
>>> On Jul 26, 2016, at 7:39 AM, Wangda Tan <wheeleast@gmail.com> wrote:
>>>
>>> lets try to use both jdiff and the new tool and compare results because
>>> this is the first time with the new tool.
>>>
>>> Appreciate your time to help us about this effort!
>>>
>>>
>>>
>>
>

Mime
  • Unnamed multipart/alternative (inline, None, 0 bytes)
View raw message