Author: bago
Date: Wed Dec 21 14:47:25 2011
New Revision: 1221748
URL: http://svn.apache.org/viewvc?rev=1221748&view=rev
Log:
An attempt to refactor AbstractProtocolTransport to be thread safe. I moved
back to standard synchronization as we only have max 2 threads competing for
the queue so it d
This makes sense! Thanks for taking care. I will cut a new release now that
this is fixed. I think I will call it 1.6.0-RC1 as we are really close ;)
Bye,
Norman
--
Norman Maurer
Am Mittwoch, 21. Dezember 2011 um 15:47 schrieb b...@apache.org:
> Author: bago
> Date: Wed Dec 21 14:47:25 2011
Hi Stephano,
Opening the discussion to learn more :)
- Why are you considering that 2 threads is a criteria to use standard
synchronization rather than some atomic fields.
- I can understand you replace a concurrent by a non-concurrent queue.
However, you now have a blocking queue. Is there
Hi Eric,
comments inside
Am 24.12.2011 um 10:05 schrieb Eric Charles :
> Hi Stephano,
>
> Opening the discussion to learn more :)
>
> - Why are you considering that 2 threads is a criteria to use standard
> synchronization rather than some atomic fields.
>
If you only have a small count
2011/12/24 Eric Charles :
> Hi Stephano,
>
> Opening the discussion to learn more :)
>
> - Why are you considering that 2 threads is a criteria to use standard
> synchronization rather than some atomic fields.
It is a criteria to not use the ConcurrentLinkedQueue that is a
structure thought to han
Hi Norman,
Thx for inputs. comment/confirmation inside.
Eric
On 24/12/11 11:53, Norman Maurer wrote:
Hi Eric,
comments inside
Am 24.12.2011 um 10:05 schrieb Eric Charles:
Hi Stephano,
Opening the discussion to learn more :)
- Why are you considering that 2 threads is a criteria to use
Hi Stephano, Thx for the inputs. I'm fine to have volatile fields and to
use them is synchronized blocks when needed, and out-of synchronized
blocks when possible.
I still have to find the few hours to better understand the usage and
context.
Maybe you can give me a hint about about the 2 us
2011/12/24 Eric Charles :
> Maybe you can give me a hint about about the 2 user threads?
The main thread is the one that write new "reponses". It is the
protocol thread and calls the writeResponse method each time a new
reponse is available.
If the response is a FutureResponse then the whole thing
Hi Eric,
Its Stefano not Stephano ;)
Comments inside...
Am Samstag, 24. Dezember 2011 schrieb Eric Charles :
> Hi Stephano, Thx for the inputs. I'm fine to have volatile fields and to
use them is synchronized blocks when needed, and out-of synchronized blocks
when possible.
>
> I still have to f
Hi there,
And again See inside,,
Am Samstag, 24. Dezember 2011 schrieb Eric Charles :
> Hi Norman,
> Thx for inputs. comment/confirmation inside.
> Eric
>
> On 24/12/11 11:53, Norman Maurer wrote:
>>
>> Hi Eric,
>>
>> comments inside
>>
>> Am 24.12.2011 um 10:05 schrieb Eric Charles:
>>
>>> H
On 24/12/11 16:31, Norman Maurer wrote:
- I can understand you replace a concurrent by a non-concurrent queue.
However, you now have a blocking queue. Is there an impact due to this
blocking aspect?
Nope there is not as we not use the blocking methods. We could even
replace it with a LinkedLi
Sorry Stefano :)
Thx Norman, I finally found the AbstractProtocolTransportTest (test
source folder was not in my build path.
Eric
On 24/12/11 16:28, Norman Maurer wrote:
Hi Eric,
Its Stefano not Stephano ;)
Comments inside...
Am Samstag, 24. Dezember 2011 schrieb Eric Charles:
Hi Stepha
12 matches
Mail list logo