[PATCH][IcedTea6] Fix Backport from S6657673
Elliott Baron
ebaron at redhat.com
Mon Apr 22 09:01:11 PDT 2013
On 04/22/2013 11:59 AM, Andrew Hughes wrote:
>
> ----- Original Message -----
>> On 04/22/2013 07:03 AM, Andrew Hughes wrote:
>>> ----- Original Message -----
>>>> Hi Andrew,
>>>>
>>>> On 04/19/2013 02:59 PM, Andrew Hughes wrote:
>>>>> ----- Original Message -----
>>>>>> Hi,
>>>>>>
>>>>>> This patch improves our backport of the Java 7 S6657673 security fix.
>>>>>> This fixes a problem where
>>>>>> javax.xml.parsers.DocumentBuilderFactory.newInstance would fail due to
>>>>>> package access restrictions on Xerces.
>>>>>> Okay to push?
>>>>>>
>>>>> No. If you're backporting patches, as has been said before, they should
>>>>> be in the openjdk directory named with the bug ID so they can be traced.
>>>>> This is hard for me to read now, never mind for long term maintenance.
>>>>>
>>>>>> Thanks,
>>>>>> Elliott
>>>>>>
>>>> There were actually a couple of patches I backported, and they were just
>>>> partial backports. I didn't think it was suitable to classify it as a
>>>> proper OpenJDK patch. The commit policy says only direct backports
>>>> should be placed in the openjdk subdirectory.
>>>>
>>>> Thanks,
>>>> Elliott
>>>>
>>> Can you provide more details on:
>>>
>>> * The repository these changes came from and which changesets
>> From the ChangeLog:
>>> * patches/security/20130416/6657673-jaxp-backport-factoryfinder.patch:
>>> Backported {parser,transform}.FactoryFinder fixes
>>> from jdk7u-dev changesets: 4a61ac055189 & 38d4d23d167c.
> Yes, I've seen this. Which repository?
http://hg.openjdk.java.net/jdk7u/jdk7u-dev/jaxp
>
>>> * What subset of the patches were used?
>> The first changeset 4a61ac055189: Adding the parameter useBSClassLoader
>> to getProviderClass and newInstance. Other portions of the patch that
>> updated the usage of these two methods to include useBSClassLoader were
>> also backported.
>> The second changeset 38d4d23d167c: Is a fixed backport of S6657673 for
>> these two FactoryFinder classes that actually sets useBSClassLoader.
>>
> Ok can we at least have these as two separate patches? I need to know
> what's going on if we're ever going to upstream this.
Alright, I will split them up.
Thanks,
Elliott
More information about the distro-pkg-dev
mailing list