[7u10] Request for approval for CR 7188852 - Move implementation of De/Inflater.getBytesRead/Writtten() to java from native
Seán Coffey
sean.coffey at oracle.com
Wed Sep 19 12:01:32 PDT 2012
On 19/09/2012 18:22, Andrew Hughes wrote:
> Sure, http://cr.openjdk.java.net/~andrew/zlib/webrev.01/ is the combined webrev.
> http://hg.openjdk.java.net/jdk8/jdk8/jdk/rev/35e024c6a62c is the original
> jdk8 changeset for 7110151.
Thanks - let's get approval from Alan/Sherman then. I think Sherman may
be OOTO at the moment. Once they're ok, I'll log a request for you to
integrate into 7u10 forests (once available).
> Well, I did propose it prior to stage 2. We're weren't given much notice that
> was going to happen (<1 week).
Yes, dates may not have been too clear around code freeze. I had
indicated that phase 2 process would begin in mid September though!
On the subject of backports, are you still planning on porting 7192804
to 7u ?
regards,
Sean.
>
> The reviewers were CCed on my original e-mail.
>
>> regards,
>> Sean.
>>
>> On 19/09/12 01:38, Andrew Hughes wrote:
>>> The MacOS X changes have added an option to jdk7u to use the system
>>> zlib library
>>> via setting SYSTEM_ZLIB=true. However, at present, this is broken
>>> on 64-bit GNU/Linux
>>> platforms, such that setting SYSTEM_ZLIB=true causes a build
>>> failure.
>>>
>>> I've already pushed a fix for this (7110151) to jdk8, which
>>> required a pre-requisite
>>> fix (7188852) to remove local changes to the zlib library. This is
>>> the first of two
>>> requests to get these fixes into 7u.
>>>
>>> 7188852, the subject of this request, moves the tracking of bytes
>>> read/written into
>>> Java code from native code. Standard system zlib tracks these
>>> values, but has an issue
>>> with possible overflow when the size of the zip goes over 4GB, such
>>> that the developers
>>> recommend that the application provides its own tracking:
>>> http://zlib.net/zlib_faq.html#faq32
>>> At present, 7u has a fix applied to the in-tree zlib code to change
>>> the type of these
>>> fields. 7188852 removes this fix and instead implements tracking
>>> in the Java code.
>>>
>>> Bug report:
>>> http://bugs.sun.com/bugdatabase/view_bug.do?bug_id=7188852
>>> Public review:
>>> http://mail.openjdk.java.net/pipermail/core-libs-dev/2012-August/011011.html
>>> Webrev: http://cr.openjdk.java.net/~andrew/7188852/webrev.01/
>>>
>>> The webrev is nearly identical to the original jdk8 changeset:
>>>
>>> http://hg.openjdk.java.net/jdk8/jdk8/jdk/rev/1468b0af0d06
>>>
>>> The only necessary changes were to fix the paths to the zlib code
>>> (7u has 1.2.3, 8 has 1.2.5)
>>> and to apply the changes to Changelog_java & inflate.c manually, as
>>> the context differed slightly.
>>> The 7u webrev builds and the included test case passes.
>>>
>>> Ok for 7u10?
>>>
>>> Thanks,
>>
More information about the jdk7u-dev
mailing list