Hi,

please don't shot the messenger.
I found what I consider to be a critical regression in the multi transaction
issue (ZK-965).
Marshalling errors may not be handled correctly since then.
Please see ZOOKEEPER-1246 for details.

Regards,

Thomas Koch, http://www.koch.ro

No worries, we don't shoot messengers we club them. ;-)

It certainly looks like a serious issue (hygiene issues aside).

Ted/Mahadev what do you think?

Patrick

On Tue, Oct 25, 2011 at 3:08 AM, Thomas Koch <[EMAIL PROTECTED]> wrote:
> Hi,
>
> please don't shot the messenger.
> I found what I consider to be a critical regression in the multi transaction
> issue (ZK-965).
> Marshalling errors may not be handled correctly since then.
> Please see ZOOKEEPER-1246 for details.
>
> Regards,
>
> Thomas Koch, http://www.koch.ro
>

The switch issue that was discussed earlier definitely looks wrong.

I haven't had time to look at the newer stuff.

On Tue, Oct 25, 2011 at 8:57 AM, Patrick Hunt <[EMAIL PROTECTED]> wrote:

> No worries, we don't shoot messengers we club them. ;-)
>
> It certainly looks like a serious issue (hygiene issues aside).
>
> Ted/Mahadev what do you think?
>
> Patrick
>
> On Tue, Oct 25, 2011 at 3:08 AM, Thomas Koch <[EMAIL PROTECTED]> wrote:
> > Hi,
> >
> > please don't shot the messenger.
> > I found what I consider to be a critical regression in the multi
> transaction
> > issue (ZK-965).
> > Marshalling errors may not be handled correctly since then.
> > Please see ZOOKEEPER-1246 for details.
> >
> > Regards,
> >
> > Thomas Koch, http://www.koch.ro
> >
>

Ted,
 Would you mind uploading a quick fix for the issue?

thanks
mahadev

On Tue, Oct 25, 2011 at 9:01 AM, Ted Dunning <[EMAIL PROTECTED]> wrote:
> The switch issue that was discussed earlier definitely looks wrong.
> I haven't had time to look at the newer stuff.
>
> On Tue, Oct 25, 2011 at 8:57 AM, Patrick Hunt <[EMAIL PROTECTED]> wrote:
>>
>> No worries, we don't shoot messengers we club them. ;-)
>>
>> It certainly looks like a serious issue (hygiene issues aside).
>>
>> Ted/Mahadev what do you think?
>>
>> Patrick
>>
>> On Tue, Oct 25, 2011 at 3:08 AM, Thomas Koch <[EMAIL PROTECTED]> wrote:
>> > Hi,
>> >
>> > please don't shot the messenger.
>> > I found what I consider to be a critical regression in the multi
>> > transaction
>> > issue (ZK-965).
>> > Marshalling errors may not be handled correctly since then.
>> > Please see ZOOKEEPER-1246 for details.
>> >
>> > Regards,
>> >
>> > Thomas Koch, http://www.koch.ro
>> >
>
>

Adding Marshall.

thanks
mahadev

On Tue, Oct 25, 2011 at 9:09 AM, Mahadev Konar <[EMAIL PROTECTED]> wrote:
> Ted,
>  Would you mind uploading a quick fix for the issue?
>
> thanks
> mahadev
>
> On Tue, Oct 25, 2011 at 9:01 AM, Ted Dunning <[EMAIL PROTECTED]> wrote:
>> The switch issue that was discussed earlier definitely looks wrong.
>> I haven't had time to look at the newer stuff.
>>
>> On Tue, Oct 25, 2011 at 8:57 AM, Patrick Hunt <[EMAIL PROTECTED]> wrote:
>>>
>>> No worries, we don't shoot messengers we club them. ;-)
>>>
>>> It certainly looks like a serious issue (hygiene issues aside).
>>>
>>> Ted/Mahadev what do you think?
>>>
>>> Patrick
>>>
>>> On Tue, Oct 25, 2011 at 3:08 AM, Thomas Koch <[EMAIL PROTECTED]> wrote:
>>> > Hi,
>>> >
>>> > please don't shot the messenger.
>>> > I found what I consider to be a critical regression in the multi
>>> > transaction
>>> > issue (ZK-965).
>>> > Marshalling errors may not be handled correctly since then.
>>> > Please see ZOOKEEPER-1246 for details.
>>> >
>>> > Regards,
>>> >
>>> > Thomas Koch, http://www.koch.ro
>>> >
>>
>>
>

I would love to.  My travel schedule just now is pretty heinous so I don't
know when I will get to it.

On Tue, Oct 25, 2011 at 9:09 AM, Mahadev Konar <[EMAIL PROTECTED]>wrote:

> Ted,
>  Would you mind uploading a quick fix for the issue?
>
> thanks
> mahadev
>
> On Tue, Oct 25, 2011 at 9:01 AM, Ted Dunning <[EMAIL PROTECTED]>
> wrote:
> > The switch issue that was discussed earlier definitely looks wrong.
> > I haven't had time to look at the newer stuff.
> >
> > On Tue, Oct 25, 2011 at 8:57 AM, Patrick Hunt <[EMAIL PROTECTED]> wrote:
> >>
> >> No worries, we don't shoot messengers we club them. ;-)
> >>
> >> It certainly looks like a serious issue (hygiene issues aside).
> >>
> >> Ted/Mahadev what do you think?
> >>
> >> Patrick
> >>
> >> On Tue, Oct 25, 2011 at 3:08 AM, Thomas Koch <[EMAIL PROTECTED]> wrote:
> >> > Hi,
> >> >
> >> > please don't shot the messenger.
> >> > I found what I consider to be a critical regression in the multi
> >> > transaction
> >> > issue (ZK-965).
> >> > Marshalling errors may not be handled correctly since then.
> >> > Please see ZOOKEEPER-1246 for details.
> >> >
> >> > Regards,
> >> >
> >> > Thomas Koch, http://www.koch.ro
> >> >
> >
> >
>

Ted Dunning:
> I would love to.  My travel schedule just now is pretty heinous so I don't
> know when I will get to it.
This is how I (hopefully) fixed this:
https://github.com/thkoch2001/zookeeper/blob/proposed_patches/src/java/main/org/apache/zookeeper/server/PrepRequestProcessor.java#L525

However I've several major changes made before that:
- enum for OpCode
- a class Request.Meta that holds all request meta-data: zxid, cxid, time,
type, owner, authinfo
- request lost all setXYZ methods except setException to move towards an
immutable request

Regards,

Thomas Koch, http://www.koch.ro

Thomas,
Any chance, you could make the change just on 3.4 branch without the
Enum changes and others?

thanks
mahadev

On Tue, Oct 25, 2011 at 10:09 AM, Thomas Koch <[EMAIL PROTECTED]> wrote:
> Ted Dunning:
>> I would love to.  My travel schedule just now is pretty heinous so I don't
>> know when I will get to it.
> This is how I (hopefully) fixed this:
> https://github.com/thkoch2001/zookeeper/blob/proposed_patches/src/java/main/org/apache/zookeeper/server/PrepRequestProcessor.java#L525
>
> However I've several major changes made before that:
> - enum for OpCode
> - a class Request.Meta that holds all request meta-data: zxid, cxid, time,
> type, owner, authinfo
> - request lost all setXYZ methods except setException to move towards an
> immutable request
>
> Regards,
>
> Thomas Koch, http://www.koch.ro
>

Do we have a test that exercises this code path? Any commit we should
really include something that verifies the change.

On Tue, Oct 25, 2011 at 10:12 AM, Mahadev Konar <[EMAIL PROTECTED]> wrote:
> Thomas,
> Any chance, you could make the change just on 3.4 branch without the
> Enum changes and others?
>
> thanks
> mahadev
>
> On Tue, Oct 25, 2011 at 10:09 AM, Thomas Koch <[EMAIL PROTECTED]> wrote:
>> Ted Dunning:
>>> I would love to.  My travel schedule just now is pretty heinous so I don't
>>> know when I will get to it.
>> This is how I (hopefully) fixed this:
>> https://github.com/thkoch2001/zookeeper/blob/proposed_patches/src/java/main/org/apache/zookeeper/server/PrepRequestProcessor.java#L525
>>
>> However I've several major changes made before that:
>> - enum for OpCode
>> - a class Request.Meta that holds all request meta-data: zxid, cxid, time,
>> type, owner, authinfo
>> - request lost all setXYZ methods except setException to move towards an
>> immutable request
>>
>> Regards,
>>
>> Thomas Koch, http://www.koch.ro
>>
>

I think mockito should be useful here. Our code needs to be a little
cleaner to be able to test these with "real" unit tests :).

thanks
mahadev

On Tue, Oct 25, 2011 at 10:14 AM, Patrick Hunt <[EMAIL PROTECTED]> wrote:
> Do we have a test that exercises this code path? Any commit we should
> really include something that verifies the change.
>
> On Tue, Oct 25, 2011 at 10:12 AM, Mahadev Konar <[EMAIL PROTECTED]> wrote:
>> Thomas,
>> Any chance, you could make the change just on 3.4 branch without the
>> Enum changes and others?
>>
>> thanks
>> mahadev
>>
>> On Tue, Oct 25, 2011 at 10:09 AM, Thomas Koch <[EMAIL PROTECTED]> wrote:
>>> Ted Dunning:
>>>> I would love to.  My travel schedule just now is pretty heinous so I don't
>>>> know when I will get to it.
>>> This is how I (hopefully) fixed this:
>>> https://github.com/thkoch2001/zookeeper/blob/proposed_patches/src/java/main/org/apache/zookeeper/server/PrepRequestProcessor.java#L525
>>>
>>> However I've several major changes made before that:
>>> - enum for OpCode
>>> - a class Request.Meta that holds all request meta-data: zxid, cxid, time,
>>> type, owner, authinfo
>>> - request lost all setXYZ methods except setException to move towards an
>>> immutable request
>>>
>>> Regards,
>>>
>>> Thomas Koch, http://www.koch.ro
>>>
>>
>

I'm happy to take a look at this and post progress. I'm in a hospital
waiting for a baby to be born so I will be on and off but I'll post
what I get through.

C

On Tue, Oct 25, 2011 at 1:16 PM, Mahadev Konar <[EMAIL PROTECTED]> wrote:
> I think mockito should be useful here. Our code needs to be a little
> cleaner to be able to test these with "real" unit tests :).
>
> thanks
> mahadev
>
> On Tue, Oct 25, 2011 at 10:14 AM, Patrick Hunt <[EMAIL PROTECTED]> wrote:
>> Do we have a test that exercises this code path? Any commit we should
>> really include something that verifies the change.
>>
>> On Tue, Oct 25, 2011 at 10:12 AM, Mahadev Konar <[EMAIL PROTECTED]> wrote:
>>> Thomas,
>>> Any chance, you could make the change just on 3.4 branch without the
>>> Enum changes and others?
>>>
>>> thanks
>>> mahadev
>>>
>>> On Tue, Oct 25, 2011 at 10:09 AM, Thomas Koch <[EMAIL PROTECTED]> wrote:
>>>> Ted Dunning:
>>>>> I would love to.  My travel schedule just now is pretty heinous so I don't
>>>>> know when I will get to it.
>>>> This is how I (hopefully) fixed this:
>>>> https://github.com/thkoch2001/zookeeper/blob/proposed_patches/src/java/main/org/apache/zookeeper/server/PrepRequestProcessor.java#L525
>>>>
>>>> However I've several major changes made before that:
>>>> - enum for OpCode
>>>> - a class Request.Meta that holds all request meta-data: zxid, cxid, time,
>>>> type, owner, authinfo
>>>> - request lost all setXYZ methods except setException to move towards an
>>>> immutable request
>>>>
>>>> Regards,
>>>>
>>>> Thomas Koch, http://www.koch.ro
>>>>
>>>
>>
>

I have a fix and a test that shows the error. Will get a patch together soon.

On Tue, Oct 25, 2011 at 3:20 PM, Camille Fournier <[EMAIL PROTECTED]> wrote:
> I'm happy to take a look at this and post progress. I'm in a hospital
> waiting for a baby to be born so I will be on and off but I'll post
> what I get through.
>
> C
>
> On Tue, Oct 25, 2011 at 1:16 PM, Mahadev Konar <[EMAIL PROTECTED]> wrote:
>> I think mockito should be useful here. Our code needs to be a little
>> cleaner to be able to test these with "real" unit tests :).
>>
>> thanks
>> mahadev
>>
>> On Tue, Oct 25, 2011 at 10:14 AM, Patrick Hunt <[EMAIL PROTECTED]> wrote:
>>> Do we have a test that exercises this code path? Any commit we should
>>> really include something that verifies the change.
>>>
>>> On Tue, Oct 25, 2011 at 10:12 AM, Mahadev Konar <[EMAIL PROTECTED]> wrote:
>>>> Thomas,
>>>> Any chance, you could make the change just on 3.4 branch without the
>>>> Enum changes and others?
>>>>
>>>> thanks
>>>> mahadev
>>>>
>>>> On Tue, Oct 25, 2011 at 10:09 AM, Thomas Koch <[EMAIL PROTECTED]> wrote:
>>>>> Ted Dunning:
>>>>>> I would love to.  My travel schedule just now is pretty heinous so I don't
>>>>>> know when I will get to it.
>>>>> This is how I (hopefully) fixed this:
>>>>> https://github.com/thkoch2001/zookeeper/blob/proposed_patches/src/java/main/org/apache/zookeeper/server/PrepRequestProcessor.java#L525
>>>>>
>>>>> However I've several major changes made before that:
>>>>> - enum for OpCode
>>>>> - a class Request.Meta that holds all request meta-data: zxid, cxid, time,
>>>>> type, owner, authinfo
>>>>> - request lost all setXYZ methods except setException to move towards an
>>>>> immutable request
>>>>>
>>>>> Regards,
>>>>>
>>>>> Thomas Koch, http://www.koch.ro
>>>>>
>>>>
>>>
>>
>

Thanks a lot Camille.

mahadev
On Tue, Oct 25, 2011 at 1:42 PM, Camille Fournier <[EMAIL PROTECTED]> wrote:
> I have a fix and a test that shows the error. Will get a patch together soon.
>
> On Tue, Oct 25, 2011 at 3:20 PM, Camille Fournier <[EMAIL PROTECTED]> wrote:
>> I'm happy to take a look at this and post progress. I'm in a hospital
>> waiting for a baby to be born so I will be on and off but I'll post
>> what I get through.
>>
>> C
>>
>> On Tue, Oct 25, 2011 at 1:16 PM, Mahadev Konar <[EMAIL PROTECTED]> wrote:
>>> I think mockito should be useful here. Our code needs to be a little
>>> cleaner to be able to test these with "real" unit tests :).
>>>
>>> thanks
>>> mahadev
>>>
>>> On Tue, Oct 25, 2011 at 10:14 AM, Patrick Hunt <[EMAIL PROTECTED]> wrote:
>>>> Do we have a test that exercises this code path? Any commit we should
>>>> really include something that verifies the change.
>>>>
>>>> On Tue, Oct 25, 2011 at 10:12 AM, Mahadev Konar <[EMAIL PROTECTED]> wrote:
>>>>> Thomas,
>>>>> Any chance, you could make the change just on 3.4 branch without the
>>>>> Enum changes and others?
>>>>>
>>>>> thanks
>>>>> mahadev
>>>>>
>>>>> On Tue, Oct 25, 2011 at 10:09 AM, Thomas Koch <[EMAIL PROTECTED]> wrote:
>>>>>> Ted Dunning:
>>>>>>> I would love to.  My travel schedule just now is pretty heinous so I don't
>>>>>>> know when I will get to it.
>>>>>> This is how I (hopefully) fixed this:
>>>>>> https://github.com/thkoch2001/zookeeper/blob/proposed_patches/src/java/main/org/apache/zookeeper/server/PrepRequestProcessor.java#L525
>>>>>>
>>>>>> However I've several major changes made before that:
>>>>>> - enum for OpCode
>>>>>> - a class Request.Meta that holds all request meta-data: zxid, cxid, time,
>>>>>> type, owner, authinfo
>>>>>> - request lost all setXYZ methods except setException to move towards an
>>>>>> immutable request
>>>>>>
>>>>>> Regards,
>>>>>>
>>>>>> Thomas Koch, http://www.koch.ro
>>>>>>
>>>>>
>>>>
>>>
>>
>

I won't reasonably have time to review the patch and test the RC0 at
scale until thursday/friday/weekend.

Mahadev what do you want to do - should we continue to beat on RC0 and
you cut a second release candidate over the weekend? Or are you
planning to cut a new RC (after fix(es) go in) on Monday? Earlier?
What would you like to do?

Patrick

On Tue, Oct 25, 2011 at 2:04 PM, Mahadev Konar <[EMAIL PROTECTED]> wrote:
> Thanks a lot Camille.
>
> mahadev
>
>
> On Tue, Oct 25, 2011 at 1:42 PM, Camille Fournier <[EMAIL PROTECTED]> wrote:
>> I have a fix and a test that shows the error. Will get a patch together soon.
>>
>> On Tue, Oct 25, 2011 at 3:20 PM, Camille Fournier <[EMAIL PROTECTED]> wrote:
>>> I'm happy to take a look at this and post progress. I'm in a hospital
>>> waiting for a baby to be born so I will be on and off but I'll post
>>> what I get through.
>>>
>>> C
>>>
>>> On Tue, Oct 25, 2011 at 1:16 PM, Mahadev Konar <[EMAIL PROTECTED]> wrote:
>>>> I think mockito should be useful here. Our code needs to be a little
>>>> cleaner to be able to test these with "real" unit tests :).
>>>>
>>>> thanks
>>>> mahadev
>>>>
>>>> On Tue, Oct 25, 2011 at 10:14 AM, Patrick Hunt <[EMAIL PROTECTED]> wrote:
>>>>> Do we have a test that exercises this code path? Any commit we should
>>>>> really include something that verifies the change.
>>>>>
>>>>> On Tue, Oct 25, 2011 at 10:12 AM, Mahadev Konar <[EMAIL PROTECTED]> wrote:
>>>>>> Thomas,
>>>>>> Any chance, you could make the change just on 3.4 branch without the
>>>>>> Enum changes and others?
>>>>>>
>>>>>> thanks
>>>>>> mahadev
>>>>>>
>>>>>> On Tue, Oct 25, 2011 at 10:09 AM, Thomas Koch <[EMAIL PROTECTED]> wrote:
>>>>>>> Ted Dunning:
>>>>>>>> I would love to.  My travel schedule just now is pretty heinous so I don't
>>>>>>>> know when I will get to it.
>>>>>>> This is how I (hopefully) fixed this:
>>>>>>> https://github.com/thkoch2001/zookeeper/blob/proposed_patches/src/java/main/org/apache/zookeeper/server/PrepRequestProcessor.java#L525
>>>>>>>
>>>>>>> However I've several major changes made before that:
>>>>>>> - enum for OpCode
>>>>>>> - a class Request.Meta that holds all request meta-data: zxid, cxid, time,
>>>>>>> type, owner, authinfo
>>>>>>> - request lost all setXYZ methods except setException to move towards an
>>>>>>> immutable request
>>>>>>>
>>>>>>> Regards,
>>>>>>>
>>>>>>> Thomas Koch, http://www.koch.ro
>>>>>>>
>>>>>>
>>>>>
>>>>
>>>
>>
>

Pat,
 I think we should continue to beat on RC0. I'd like to cut  a new
release candidate after the weekend. That will give everyone some
weekend time to play with it.

I'll aim for a new RC on Monday/Tuesday.
Seems reasonable?

thanks
mahadev

On Wed, Oct 26, 2011 at 4:44 PM, Patrick Hunt <[EMAIL PROTECTED]> wrote:
> I won't reasonably have time to review the patch and test the RC0 at
> scale until thursday/friday/weekend.
>
> Mahadev what do you want to do - should we continue to beat on RC0 and
> you cut a second release candidate over the weekend? Or are you
> planning to cut a new RC (after fix(es) go in) on Monday? Earlier?
> What would you like to do?
>
> Patrick
>
> On Tue, Oct 25, 2011 at 2:04 PM, Mahadev Konar <[EMAIL PROTECTED]> wrote:
>> Thanks a lot Camille.
>>
>> mahadev
>>
>>
>> On Tue, Oct 25, 2011 at 1:42 PM, Camille Fournier <[EMAIL PROTECTED]> wrote:
>>> I have a fix and a test that shows the error. Will get a patch together soon.
>>>
>>> On Tue, Oct 25, 2011 at 3:20 PM, Camille Fournier <[EMAIL PROTECTED]> wrote:
>>>> I'm happy to take a look at this and post progress. I'm in a hospital
>>>> waiting for a baby to be born so I will be on and off but I'll post
>>>> what I get through.
>>>>
>>>> C
>>>>
>>>> On Tue, Oct 25, 2011 at 1:16 PM, Mahadev Konar <[EMAIL PROTECTED]> wrote:
>>>>> I think mockito should be useful here. Our code needs to be a little
>>>>> cleaner to be able to test these with "real" unit tests :).
>>>>>
>>>>> thanks
>>>>> mahadev
>>>>>
>>>>> On Tue, Oct 25, 2011 at 10:14 AM, Patrick Hunt <[EMAIL PROTECTED]> wrote:
>>>>>> Do we have a test that exercises this code path? Any commit we should
>>>>>> really include something that verifies the change.
>>>>>>
>>>>>> On Tue, Oct 25, 2011 at 10:12 AM, Mahadev Konar <[EMAIL PROTECTED]> wrote:
>>>>>>> Thomas,
>>>>>>> Any chance, you could make the change just on 3.4 branch without the
>>>>>>> Enum changes and others?
>>>>>>>
>>>>>>> thanks
>>>>>>> mahadev
>>>>>>>
>>>>>>> On Tue, Oct 25, 2011 at 10:09 AM, Thomas Koch <[EMAIL PROTECTED]> wrote:
>>>>>>>> Ted Dunning:
>>>>>>>>> I would love to.  My travel schedule just now is pretty heinous so I don't
>>>>>>>>> know when I will get to it.
>>>>>>>> This is how I (hopefully) fixed this:
>>>>>>>> https://github.com/thkoch2001/zookeeper/blob/proposed_patches/src/java/main/org/apache/zookeeper/server/PrepRequestProcessor.java#L525
>>>>>>>>
>>>>>>>> However I've several major changes made before that:
>>>>>>>> - enum for OpCode
>>>>>>>> - a class Request.Meta that holds all request meta-data: zxid, cxid, time,
>>>>>>>> type, owner, authinfo
>>>>>>>> - request lost all setXYZ methods except setException to move towards an
>>>>>>>> immutable request
>>>>>>>>
>>>>>>>> Regards,
>>>>>>>>
>>>>>>>> Thomas Koch, http://www.koch.ro
>>>>>>>>
>>>>>>>
>>>>>>
>>>>>
>>>>
>>>
>>
>

That sounds good to me. I'm planning to do RC0 testing over the next few days.

Patrick

On Thu, Oct 27, 2011 at 9:35 AM, Mahadev Konar <[EMAIL PROTECTED]> wrote:
> Pat,
>  I think we should continue to beat on RC0. I'd like to cut  a new
> release candidate after the weekend. That will give everyone some
> weekend time to play with it.
>
> I'll aim for a new RC on Monday/Tuesday.
> Seems reasonable?
>
> thanks
> mahadev
>
> On Wed, Oct 26, 2011 at 4:44 PM, Patrick Hunt <[EMAIL PROTECTED]> wrote:
>> I won't reasonably have time to review the patch and test the RC0 at
>> scale until thursday/friday/weekend.
>>
>> Mahadev what do you want to do - should we continue to beat on RC0 and
>> you cut a second release candidate over the weekend? Or are you
>> planning to cut a new RC (after fix(es) go in) on Monday? Earlier?
>> What would you like to do?
>>
>> Patrick
>>
>> On Tue, Oct 25, 2011 at 2:04 PM, Mahadev Konar <[EMAIL PROTECTED]> wrote:
>>> Thanks a lot Camille.
>>>
>>> mahadev
>>>
>>>
>>> On Tue, Oct 25, 2011 at 1:42 PM, Camille Fournier <[EMAIL PROTECTED]> wrote:
>>>> I have a fix and a test that shows the error. Will get a patch together soon.
>>>>
>>>> On Tue, Oct 25, 2011 at 3:20 PM, Camille Fournier <[EMAIL PROTECTED]> wrote:
>>>>> I'm happy to take a look at this and post progress. I'm in a hospital
>>>>> waiting for a baby to be born so I will be on and off but I'll post
>>>>> what I get through.
>>>>>
>>>>> C
>>>>>
>>>>> On Tue, Oct 25, 2011 at 1:16 PM, Mahadev Konar <[EMAIL PROTECTED]> wrote:
>>>>>> I think mockito should be useful here. Our code needs to be a little
>>>>>> cleaner to be able to test these with "real" unit tests :).
>>>>>>
>>>>>> thanks
>>>>>> mahadev
>>>>>>
>>>>>> On Tue, Oct 25, 2011 at 10:14 AM, Patrick Hunt <[EMAIL PROTECTED]> wrote:
>>>>>>> Do we have a test that exercises this code path? Any commit we should
>>>>>>> really include something that verifies the change.
>>>>>>>
>>>>>>> On Tue, Oct 25, 2011 at 10:12 AM, Mahadev Konar <[EMAIL PROTECTED]> wrote:
>>>>>>>> Thomas,
>>>>>>>> Any chance, you could make the change just on 3.4 branch without the
>>>>>>>> Enum changes and others?
>>>>>>>>
>>>>>>>> thanks
>>>>>>>> mahadev
>>>>>>>>
>>>>>>>> On Tue, Oct 25, 2011 at 10:09 AM, Thomas Koch <[EMAIL PROTECTED]> wrote:
>>>>>>>>> Ted Dunning:
>>>>>>>>>> I would love to.  My travel schedule just now is pretty heinous so I don't
>>>>>>>>>> know when I will get to it.
>>>>>>>>> This is how I (hopefully) fixed this:
>>>>>>>>> https://github.com/thkoch2001/zookeeper/blob/proposed_patches/src/java/main/org/apache/zookeeper/server/PrepRequestProcessor.java#L525
>>>>>>>>>
>>>>>>>>> However I've several major changes made before that:
>>>>>>>>> - enum for OpCode
>>>>>>>>> - a class Request.Meta that holds all request meta-data: zxid, cxid, time,
>>>>>>>>> type, owner, authinfo
>>>>>>>>> - request lost all setXYZ methods except setException to move towards an
>>>>>>>>> immutable request
>>>>>>>>>
>>>>>>>>> Regards,
>>>>>>>>>
>>>>>>>>> Thomas Koch, http://www.koch.ro
>>>>>>>>>
>>>>>>>>
>>>>>>>
>>>>>>
>>>>>
>>>>
>>>
>>
>