Re: RFR: 8186334: JarFile throws ArrayIndexOutOfBoundsException when the manifest contains certain characters

[Claes Redestad]

I think it makes a lot of sense to get this into a 9.x update, and I'll 
work through the details as soon as there's a 9 update forest available 
to commit backports to.


/Claes

On 2017-08-21 20:05, Jonathan Bluett-Duncan wrote:
Would it not be viable to merge this into Java 9 in the foreseeable 
future for one of its inevitable bug fix releases? I admit I'm 
unfamiliar with the whole process, so I realise my question may be 
naive or that the answer is considered "common knowledge".


Cheers,
Jonathan

On 21 August 2017 at 18:41, Claes Redestad > wrote:




On 2017-08-21 19:15, Paul Sandoz wrote:

On 21 Aug 2017, at 02:53, Claes Redestad
> wrote:

Hi,

this patch addresses an unfortunate regression where
backtick characters
in a manifest can cause an AIOOBE.

Webrev:
http://cr.openjdk.java.net/~redestad/8186334/jdk.00/

Bug: https://bugs.openjdk.java.net/browse/JDK-8186334


Basically an off-by-one issue during certain steps in the
search algorithm,
meaning it is context dependent whether a backtick will
trip on this issue
or not.

Ooops, looks good.


Thanks for reviewing!


I see a workaround has been pushed to Jackson, which reduced
my urge to suggest a nine respin. But… being slightly paranoid
testing JarFile on a local maven central mirror would give us
a better sense of the impact.


I think the 9 train has left the station even for issues of this
severity. And seeing as there are somewhat straightforward
workarounds I guess we'll have to live with my mistakes. Sorry!

/Claes

Re: RFR: 8186334: JarFile throws ArrayIndexOutOfBoundsException when the manifest contains certain characters

[Brent Christian]

Hi, Claes

Fix looks good, test looks good.

Thanks,
-Brent
On 08/21/2017 02:53 AM, Claes Redestad wrote:

Hi,

this patch addresses an unfortunate regression where backtick characters
in a manifest can cause an AIOOBE.

Webrev: http://cr.openjdk.java.net/~redestad/8186334/jdk.00/
Bug: https://bugs.openjdk.java.net/browse/JDK-8186334

Basically an off-by-one issue during certain steps in the search algorithm,
meaning it is context dependent whether a backtick will trip on this issue
or not.

/Claes

Re: RFR: 8186334: JarFile throws ArrayIndexOutOfBoundsException when the manifest contains certain characters

[Jonathan Bluett-Duncan]

Would it not be viable to merge this into Java 9 in the foreseeable future
for one of its inevitable bug fix releases? I admit I'm unfamiliar with the
whole process, so I realise my question may be naive or that the answer is
considered "common knowledge".

Cheers,
Jonathan

On 21 August 2017 at 18:41, Claes Redestad 
wrote:

>
>
> On 2017-08-21 19:15, Paul Sandoz wrote:
>
>> On 21 Aug 2017, at 02:53, Claes Redestad 
>>> wrote:
>>>
>>> Hi,
>>>
>>> this patch addresses an unfortunate regression where backtick characters
>>> in a manifest can cause an AIOOBE.
>>>
>>> Webrev: http://cr.openjdk.java.net/~redestad/8186334/jdk.00/
>>> Bug: https://bugs.openjdk.java.net/browse/JDK-8186334
>>>
>>> Basically an off-by-one issue during certain steps in the search
>>> algorithm,
>>> meaning it is context dependent whether a backtick will trip on this
>>> issue
>>> or not.
>>>
>>> Ooops, looks good.
>>
>
> Thanks for reviewing!
>
>
>> I see a workaround has been pushed to Jackson, which reduced my urge to
>> suggest a nine respin. But… being slightly paranoid testing JarFile on a
>> local maven central mirror would give us a better sense of the impact.
>>
>
> I think the 9 train has left the station even for issues of this severity.
> And seeing as there are somewhat straightforward workarounds I guess we'll
> have to live with my mistakes. Sorry!
>
> /Claes
>

Re: RFR: 8186334: JarFile throws ArrayIndexOutOfBoundsException when the manifest contains certain characters

[Claes Redestad]

On 2017-08-21 19:15, Paul Sandoz wrote:

On 21 Aug 2017, at 02:53, Claes Redestad  wrote:

Hi,

this patch addresses an unfortunate regression where backtick characters
in a manifest can cause an AIOOBE.

Webrev: http://cr.openjdk.java.net/~redestad/8186334/jdk.00/
Bug: https://bugs.openjdk.java.net/browse/JDK-8186334

Basically an off-by-one issue during certain steps in the search algorithm,
meaning it is context dependent whether a backtick will trip on this issue
or not.


Ooops, looks good.


Thanks for reviewing!



I see a workaround has been pushed to Jackson, which reduced my urge to suggest 
a nine respin. But… being slightly paranoid testing JarFile on a local maven 
central mirror would give us a better sense of the impact.


I think the 9 train has left the station even for issues of this 
severity. And seeing as there are somewhat straightforward workarounds I 
guess we'll have to live with my mistakes. Sorry!


/Claes

Re: RFR: 8186334: JarFile throws ArrayIndexOutOfBoundsException when the manifest contains certain characters

[Paul Sandoz]

> On 21 Aug 2017, at 02:53, Claes Redestad  wrote:
> 
> Hi,
> 
> this patch addresses an unfortunate regression where backtick characters
> in a manifest can cause an AIOOBE.
> 
> Webrev: http://cr.openjdk.java.net/~redestad/8186334/jdk.00/
> Bug: https://bugs.openjdk.java.net/browse/JDK-8186334
> 
> Basically an off-by-one issue during certain steps in the search algorithm,
> meaning it is context dependent whether a backtick will trip on this issue
> or not.
> 

Ooops, looks good.

I see a workaround has been pushed to Jackson, which reduced my urge to suggest 
a nine respin. But… being slightly paranoid testing JarFile on a local maven 
central mirror would give us a better sense of the impact.

Paul.

> /Claes

RFR: 8186334: JarFile throws ArrayIndexOutOfBoundsException when the manifest contains certain characters

[Claes Redestad]

Hi,

this patch addresses an unfortunate regression where backtick characters
in a manifest can cause an AIOOBE.

Webrev: http://cr.openjdk.java.net/~redestad/8186334/jdk.00/
Bug: https://bugs.openjdk.java.net/browse/JDK-8186334

Basically an off-by-one issue during certain steps in the search algorithm,
meaning it is context dependent whether a backtick will trip on this issue
or not.

/Claes