RFC: nashorn sources are missing in src.zip

[David Holmes]

Hi Jiri,

On 4/01/2017 1:11 AM, Jiri Vanek wrote:
> On 01/03/2017 01:43 PM, Erik Joelsson wrote:
>> Hello Jiri,
>>
>> From a pure build point of view, the change looks ok, but I don't
>> think we are allowed to accept
>> patches not hosted on openjdk infrastructure.
>
> Those patches were already accepted. And the policy "patch first, member
> later"  is actually enforcing patch elsewhere.
>
> I'm already in author group (jdk9 project only) so I should have place
> on java.net, so if this is blocker, it can be fixed.

Yes please fix, it is a requirement for contribution acceptance.

>>
>> The suitability of the change is a topic for jdk8u-dev (added). This
>> should be considered an
>
> Thanx!
>
>> enhancement though. The restrictive inclusion of packages in JDK 8 and
>> older was intentional and
>> changing it in JDK 9 was definitely an enhancement. (Note that JDK 9
>> recently changed the src.zip
>> again)
>>
>> One could argue that your original change to just include the jdk
>> package is a bug fix since the
>> intention of the original filtering was to just include public APIs.
>
>
> Its up to you and build-dev to decide this.
> If nashorn-only change will be accepted, Then I will very soon post also
> zipfs  patch, as it is very similar.

This is an enhancement and needs to follow the process described here:

http://openjdk.java.net/projects/jdk8u/enhancement-template.html

that process itself is somewhat hard to find but referenced from Rule 3 
here:

http://openjdk.java.net/projects/jdk8u/groundrules.html

Thanks,
David

> If not, then it will become just another distro-only (fedora/rhel)
> patch, as those sources *are* missing (if not all then nashorn definitely)
>
> Thanx a lot!
>
>   All the best,
>     J.
>>
>> /Erik
>>
>> On 2017-01-03 13:12, Jiri Vanek wrote:
>>> Hello!
>>>
>>> Thank you very much for kind answer and additional resources.
>>> There is modified backport of patch you mentioned:
>>> https://jvanek.fedorapeople.org/oracle/jdk8/webrevs/nashornMissingInSrcZip/v2/webrev/
>>>
>>>
>>> Except the backport itself, I had to add also nashorn sources (as
>>> jdk9 is handling this a bit
>>> differently).
>>> At the end I added also sources for  zipfs, as jdk9 have them too,
>>> and they were the last missing
>>> bits as far as I saw.
>>>
>>> The size of src.zip grown from 26 to 52MB (still less then 62 of jdk9).
>>>
>>> I think the change i suitable for 8u, as it is bug in truth. I found
>>> it when I faced strange
>>> behaviour in rhino and wonted to step in (with src.zip linked to
>>> ide)...and... nothing. So it was
>>> not nice surprise.
>>>
>>> Tahnx again!
>>>   J.
>>>
>>> On 01/03/2017 09:03 AM, Erik Joelsson wrote:
>>>> Hello Jiri,
>>>>
>>>> Please see related issue [1] which was included early in JDK 9 but
>>>> the backport was never done. I
>>>> don't personally mind including more sources in src.zip for OpenJDK,
>>>> but instead of adding them
>>>> piecemeal I would argue for reopening the backport. I don't know if
>>>> such a change is suitable for 8u
>>>> at this point though. The size of src.zip increases quite
>>>> dramatically. However, it should be noted
>>>> that it's only included in the JDK image and not the JRE.
>>>>
>>>> For anyone unsure of the role of src.zip, it's only use is for IDEs
>>>> to see the JDK sources to aid
>>>> debugging of other applications.
>>>>
>>>> /Erik
>>>>
>>>> [1] https://bugs.openjdk.java.net/browse/JDK-8044235
>>>>
>>>> On 2017-01-02 17:41, Jiri Vanek wrote:
>>>>> https://jvanek.fedorapeople.org/oracle/jdk8/webrevs/nashornMissingInSrcZip/v1/webrev/
>>>>>
>>>>>
>>>>> Hello good people!
>>>>>
>>>>> I have found, that nashorn sources are missing in src.zip.
>>>>> It is easily fixable by webrev above.
>>>>>
>>>>> If you will think about it as "private api",  jdk9 *have* nashorn
>>>>> module sources included.
>>>>>
>>>>> Actually more sun* sources are missing. Or even more sources which
>>>>> result to ext directory. Imho
>>>>> they should be included too. I will prepare patches for each set
>>>>> separately, if it will be agreed.
>>>>>
>>>>> I know that sun* are somehow deprecated. But it is sad when you can
>>>>> not easily debug inside the
>>>>> JDK. The absence of theirs in src.zip is not the cure.
>>>>>
>>>>> Thanx!
>>>>>   J.
>>>>
>>>
>>
>