You are viewing a plain text version of this content. The canonical link for it is here.
Posted to dev@jmeter.apache.org by sebb <se...@gmail.com> on 2016/01/04 17:23:47 UTC

Re: Drop Commons-HttpClient 3.1

On 27 December 2015 at 17:39, Philippe Mouawad
<ph...@gmail.com> wrote:
> Hello,
> I think it is time now that we definitely drop HC3.1 support in JMeter.
> It hardens maintainability and hides some issues in test cases like Bug
> 58773 and previously TestCookieManager tests.
>
> I think also that giving too much choice to users hardens the learning path.
>
> I suggest we start by deprecating in 2.14 the following components:
>
>    - SoapSampler, we should I think deprecate it already in 2.14 as the
>    Soap Template covers its features in a more efficient way
>    - HttpHC3Impl in documentation
>    - HC3CookieHandler in documentation

OK

> And in next version drop:
>
>    - HTTPHC3Impl
>    - SoapSampler
>    - HTTPSampler2 used by SoapSampler
>    - HC3CookieHandler
>    - HttpHC3Impl
>    - LoopbackHttpClientSocketFactory
>    - SlowHttpClientSocketFactory

OK

>
> --
> Regards
> Philippe

Re: Drop Commons-HttpClient 3.1

Posted by Milamber <mi...@apache.org>.

On 14/02/2017 20:57, Philippe Mouawad wrote:
> On Tue, Feb 14, 2017 at 9:49 PM, Milamber <mi...@apache.org> wrote:
>
>> IMHO, wait 72h to have some feedback or (if no feedback: a lazy consensus)
>> seems good (72h is the traditional delay is ASF)
>>
> For me it's not an official vote, so the 72 h is not necessary right ?

That is not a vote (or a technical vote) but you ask for feedback with a 
delay (tomorrow and after 24h), so I said the traditional delay is 72h, 
perhaps that would be good to wait directly 72h.

> although we've waited more than a year :-)
> But I'll wait :-)
>
>
>
>> For my opinion about the removing the HC 3.1: yes and no...
>>
>> The HC4 (in default config) don't react like HC3.1 (for me), I need to set
>> up this props to my load tests:
>> http.connection.stalecheck$Boolean=true
>>
> httpclient4.retrycount=1
> Did you try last nightly build ?

Yes, but always with theses properties.

> This should not be necessary anymore.
> If it's still necessary then it means there's a problem.

I will try without these props to see if the issue (fake error) continue 
to occur.

>
> For record, I proposed in the past to set retry to 2 to act like browser,
> but the discussion led to saying that it could hide issues, which is true,
> but does not mimic Browser behaviour :-)
>
>
>
>> Sometimes I prefer switch to HC3.1 to avoid this setup...
>>
>
>
>> Yes to change, because HC4 is the future even if some behaviors are
>> different (and I would like the HTTP/2)
>> No (a little no) because I'm lazy ;)
>>
>>
>>
>> On 14/02/2017 19:51, Philippe Mouawad wrote:
>>
>>> Hi,
>>> I'll wait another 24h.
>>> Feedback welcome
>>> Regards
>>>
>>>
>>> On Mon, Feb 13, 2017 at 9:05 PM, Philippe Mouawad <
>>> philippe.mouawad@gmail.com> wrote:
>>>
>>> Hello,
>>>> Unless there is NOGO on this, I'll be committing the patch provided
>>>> tomorrow on issue:
>>>>
>>>>      - https://bz.apache.org/bugzilla/show_bug.cgi?id=59038
>>>>
>>>> To recap, we deprecated HC3.1 in 3.0 and announced it would be removed in
>>>> 3.1 but we didn't remove it.
>>>> The only little issue I see in dropping it, is that we didn't deprecate
>>>> (by error) SOAP/XML-RPC Request:
>>>>
>>>>      - http://jmeter.apache.org/usermanual/component_reference.
>>>>      html#SOAP/XML-RPC_Request
>>>>
>>>> although we wrote:
>>>>
>>>>      - See Building a WebService Test Plan
>>>>      <http://jmeter.apache.org/usermanual/build-ws-test-plan.html> for up
>>>>
>>>>      to date way of test SOAP and REST Webservices
>>>>
>>>> But I'm personally in favor of dropping it for 3.2 as the only remaining
>>>> issue was the missing Pre-Emptive Basic Auth which has now been fixed in
>>>> HC4.
>>>>
>>>> Regards
>>>>
>>>> Philippe
>>>>
>>>> On Monday, January 4, 2016, sebb <se...@gmail.com> wrote:
>>>>
>>>> On 27 December 2015 at 17:39, Philippe Mouawad
>>>>> <ph...@gmail.com> wrote:
>>>>>
>>>>>> Hello,
>>>>>> I think it is time now that we definitely drop HC3.1 support in JMeter.
>>>>>> It hardens maintainability and hides some issues in test cases like Bug
>>>>>> 58773 and previously TestCookieManager tests.
>>>>>>
>>>>>> I think also that giving too much choice to users hardens the learning
>>>>>>
>>>>> path.
>>>>>
>>>>>> I suggest we start by deprecating in 2.14 the following components:
>>>>>>
>>>>>>      - SoapSampler, we should I think deprecate it already in 2.14 as
>>>>>> the
>>>>>>      Soap Template covers its features in a more efficient way
>>>>>>      - HttpHC3Impl in documentation
>>>>>>      - HC3CookieHandler in documentation
>>>>>>
>>>>> OK
>>>>>
>>>>> And in next version drop:
>>>>>>      - HTTPHC3Impl
>>>>>>      - SoapSampler
>>>>>>      - HTTPSampler2 used by SoapSampler
>>>>>>      - HC3CookieHandler
>>>>>>      - HttpHC3Impl
>>>>>>      - LoopbackHttpClientSocketFactory
>>>>>>      - SlowHttpClientSocketFactory
>>>>>>
>>>>> OK
>>>>>
>>>>> --
>>>>>> Regards
>>>>>> Philippe
>>>>>>
>


Re: Drop Commons-HttpClient 3.1

Posted by Philippe Mouawad <ph...@gmail.com>.
On Tue, Feb 14, 2017 at 9:49 PM, Milamber <mi...@apache.org> wrote:

>
> IMHO, wait 72h to have some feedback or (if no feedback: a lazy consensus)
> seems good (72h is the traditional delay is ASF)
>
For me it's not an official vote, so the 72 h is not necessary right ?
although we've waited more than a year :-)
But I'll wait :-)



> For my opinion about the removing the HC 3.1: yes and no...
>
> The HC4 (in default config) don't react like HC3.1 (for me), I need to set
> up this props to my load tests:
> http.connection.stalecheck$Boolean=true
>
httpclient4.retrycount=1
>

Did you try last nightly build ?
This should not be necessary anymore.
If it's still necessary then it means there's a problem.

For record, I proposed in the past to set retry to 2 to act like browser,
but the discussion led to saying that it could hide issues, which is true,
but does not mimic Browser behaviour :-)



> Sometimes I prefer switch to HC3.1 to avoid this setup...
>



>
> Yes to change, because HC4 is the future even if some behaviors are
> different (and I would like the HTTP/2)
> No (a little no) because I'm lazy ;)
>
>
>
> On 14/02/2017 19:51, Philippe Mouawad wrote:
>
>> Hi,
>> I'll wait another 24h.
>> Feedback welcome
>> Regards
>>
>>
>> On Mon, Feb 13, 2017 at 9:05 PM, Philippe Mouawad <
>> philippe.mouawad@gmail.com> wrote:
>>
>> Hello,
>>> Unless there is NOGO on this, I'll be committing the patch provided
>>> tomorrow on issue:
>>>
>>>     - https://bz.apache.org/bugzilla/show_bug.cgi?id=59038
>>>
>>> To recap, we deprecated HC3.1 in 3.0 and announced it would be removed in
>>> 3.1 but we didn't remove it.
>>> The only little issue I see in dropping it, is that we didn't deprecate
>>> (by error) SOAP/XML-RPC Request:
>>>
>>>     - http://jmeter.apache.org/usermanual/component_reference.
>>>     html#SOAP/XML-RPC_Request
>>>
>>> although we wrote:
>>>
>>>     - See Building a WebService Test Plan
>>>     <http://jmeter.apache.org/usermanual/build-ws-test-plan.html> for up
>>>
>>>     to date way of test SOAP and REST Webservices
>>>
>>> But I'm personally in favor of dropping it for 3.2 as the only remaining
>>> issue was the missing Pre-Emptive Basic Auth which has now been fixed in
>>> HC4.
>>>
>>> Regards
>>>
>>> Philippe
>>>
>>> On Monday, January 4, 2016, sebb <se...@gmail.com> wrote:
>>>
>>> On 27 December 2015 at 17:39, Philippe Mouawad
>>>> <ph...@gmail.com> wrote:
>>>>
>>>>> Hello,
>>>>> I think it is time now that we definitely drop HC3.1 support in JMeter.
>>>>> It hardens maintainability and hides some issues in test cases like Bug
>>>>> 58773 and previously TestCookieManager tests.
>>>>>
>>>>> I think also that giving too much choice to users hardens the learning
>>>>>
>>>> path.
>>>>
>>>>> I suggest we start by deprecating in 2.14 the following components:
>>>>>
>>>>>     - SoapSampler, we should I think deprecate it already in 2.14 as
>>>>> the
>>>>>     Soap Template covers its features in a more efficient way
>>>>>     - HttpHC3Impl in documentation
>>>>>     - HC3CookieHandler in documentation
>>>>>
>>>> OK
>>>>
>>>> And in next version drop:
>>>>>
>>>>>     - HTTPHC3Impl
>>>>>     - SoapSampler
>>>>>     - HTTPSampler2 used by SoapSampler
>>>>>     - HC3CookieHandler
>>>>>     - HttpHC3Impl
>>>>>     - LoopbackHttpClientSocketFactory
>>>>>     - SlowHttpClientSocketFactory
>>>>>
>>>> OK
>>>>
>>>> --
>>>>> Regards
>>>>> Philippe
>>>>>
>>>>
>>
>


-- 
Cordialement.
Philippe Mouawad.

Re: Drop Commons-HttpClient 3.1

Posted by Milamber <mi...@apache.org>.
IMHO, wait 72h to have some feedback or (if no feedback: a lazy 
consensus) seems good (72h is the traditional delay is ASF)

For my opinion about the removing the HC 3.1: yes and no...

The HC4 (in default config) don't react like HC3.1 (for me), I need to 
set up this props to my load tests:
http.connection.stalecheck$Boolean=true
httpclient4.retrycount=1

Sometimes I prefer switch to HC3.1 to avoid this setup...

Yes to change, because HC4 is the future even if some behaviors are 
different (and I would like the HTTP/2)
No (a little no) because I'm lazy ;)



On 14/02/2017 19:51, Philippe Mouawad wrote:
> Hi,
> I'll wait another 24h.
> Feedback welcome
> Regards
>
>
> On Mon, Feb 13, 2017 at 9:05 PM, Philippe Mouawad <
> philippe.mouawad@gmail.com> wrote:
>
>> Hello,
>> Unless there is NOGO on this, I'll be committing the patch provided
>> tomorrow on issue:
>>
>>     - https://bz.apache.org/bugzilla/show_bug.cgi?id=59038
>>
>> To recap, we deprecated HC3.1 in 3.0 and announced it would be removed in
>> 3.1 but we didn't remove it.
>> The only little issue I see in dropping it, is that we didn't deprecate
>> (by error) SOAP/XML-RPC Request:
>>
>>     - http://jmeter.apache.org/usermanual/component_reference.
>>     html#SOAP/XML-RPC_Request
>>
>> although we wrote:
>>
>>     - See Building a WebService Test Plan
>>     <http://jmeter.apache.org/usermanual/build-ws-test-plan.html> for up
>>     to date way of test SOAP and REST Webservices
>>
>> But I'm personally in favor of dropping it for 3.2 as the only remaining
>> issue was the missing Pre-Emptive Basic Auth which has now been fixed in
>> HC4.
>>
>> Regards
>>
>> Philippe
>>
>> On Monday, January 4, 2016, sebb <se...@gmail.com> wrote:
>>
>>> On 27 December 2015 at 17:39, Philippe Mouawad
>>> <ph...@gmail.com> wrote:
>>>> Hello,
>>>> I think it is time now that we definitely drop HC3.1 support in JMeter.
>>>> It hardens maintainability and hides some issues in test cases like Bug
>>>> 58773 and previously TestCookieManager tests.
>>>>
>>>> I think also that giving too much choice to users hardens the learning
>>> path.
>>>> I suggest we start by deprecating in 2.14 the following components:
>>>>
>>>>     - SoapSampler, we should I think deprecate it already in 2.14 as the
>>>>     Soap Template covers its features in a more efficient way
>>>>     - HttpHC3Impl in documentation
>>>>     - HC3CookieHandler in documentation
>>> OK
>>>
>>>> And in next version drop:
>>>>
>>>>     - HTTPHC3Impl
>>>>     - SoapSampler
>>>>     - HTTPSampler2 used by SoapSampler
>>>>     - HC3CookieHandler
>>>>     - HttpHC3Impl
>>>>     - LoopbackHttpClientSocketFactory
>>>>     - SlowHttpClientSocketFactory
>>> OK
>>>
>>>> --
>>>> Regards
>>>> Philippe
>


Re: Drop Commons-HttpClient 3.1

Posted by Philippe Mouawad <ph...@gmail.com>.
Hi,
I'll wait another 24h.
Feedback welcome
Regards


On Mon, Feb 13, 2017 at 9:05 PM, Philippe Mouawad <
philippe.mouawad@gmail.com> wrote:

> Hello,
> Unless there is NOGO on this, I'll be committing the patch provided
> tomorrow on issue:
>
>    - https://bz.apache.org/bugzilla/show_bug.cgi?id=59038
>
> To recap, we deprecated HC3.1 in 3.0 and announced it would be removed in
> 3.1 but we didn't remove it.
> The only little issue I see in dropping it, is that we didn't deprecate
> (by error) SOAP/XML-RPC Request:
>
>    - http://jmeter.apache.org/usermanual/component_reference.
>    html#SOAP/XML-RPC_Request
>
> although we wrote:
>
>    - See Building a WebService Test Plan
>    <http://jmeter.apache.org/usermanual/build-ws-test-plan.html> for up
>    to date way of test SOAP and REST Webservices
>
> But I'm personally in favor of dropping it for 3.2 as the only remaining
> issue was the missing Pre-Emptive Basic Auth which has now been fixed in
> HC4.
>
> Regards
>
> Philippe
>
> On Monday, January 4, 2016, sebb <se...@gmail.com> wrote:
>
>> On 27 December 2015 at 17:39, Philippe Mouawad
>> <ph...@gmail.com> wrote:
>> > Hello,
>> > I think it is time now that we definitely drop HC3.1 support in JMeter.
>> > It hardens maintainability and hides some issues in test cases like Bug
>> > 58773 and previously TestCookieManager tests.
>> >
>> > I think also that giving too much choice to users hardens the learning
>> path.
>> >
>> > I suggest we start by deprecating in 2.14 the following components:
>> >
>> >    - SoapSampler, we should I think deprecate it already in 2.14 as the
>> >    Soap Template covers its features in a more efficient way
>> >    - HttpHC3Impl in documentation
>> >    - HC3CookieHandler in documentation
>>
>> OK
>>
>> > And in next version drop:
>> >
>> >    - HTTPHC3Impl
>> >    - SoapSampler
>> >    - HTTPSampler2 used by SoapSampler
>> >    - HC3CookieHandler
>> >    - HttpHC3Impl
>> >    - LoopbackHttpClientSocketFactory
>> >    - SlowHttpClientSocketFactory
>>
>> OK
>>
>> >
>> > --
>> > Regards
>> > Philippe
>>
>


-- 
Cordialement.
Philippe Mouawad.

Re: Drop Commons-HttpClient 3.1

Posted by Philippe Mouawad <ph...@gmail.com>.
Hello,
Unless there is NOGO on this, I'll be committing the patch provided
tomorrow on issue:

   - https://bz.apache.org/bugzilla/show_bug.cgi?id=59038

To recap, we deprecated HC3.1 in 3.0 and announced it would be removed in
3.1 but we didn't remove it.
The only little issue I see in dropping it, is that we didn't deprecate (by
error) SOAP/XML-RPC Request:

   - http://jmeter.apache.org/usermanual/component_
   reference.html#SOAP/XML-RPC_Request
   <http://jmeter.apache.org/usermanual/component_reference.html#SOAP/XML-RPC_Request>

although we wrote:

   - See Building a WebService Test Plan
   <http://jmeter.apache.org/usermanual/build-ws-test-plan.html> for up to
   date way of test SOAP and REST Webservices

But I'm personally in favor of dropping it for 3.2 as the only remaining
issue was the missing Pre-Emptive Basic Auth which has now been fixed in
HC4.

Regards

Philippe

On Monday, January 4, 2016, sebb <se...@gmail.com> wrote:

> On 27 December 2015 at 17:39, Philippe Mouawad
> <ph...@gmail.com> wrote:
> > Hello,
> > I think it is time now that we definitely drop HC3.1 support in JMeter.
> > It hardens maintainability and hides some issues in test cases like Bug
> > 58773 and previously TestCookieManager tests.
> >
> > I think also that giving too much choice to users hardens the learning
> path.
> >
> > I suggest we start by deprecating in 2.14 the following components:
> >
> >    - SoapSampler, we should I think deprecate it already in 2.14 as the
> >    Soap Template covers its features in a more efficient way
> >    - HttpHC3Impl in documentation
> >    - HC3CookieHandler in documentation
>
> OK
>
> > And in next version drop:
> >
> >    - HTTPHC3Impl
> >    - SoapSampler
> >    - HTTPSampler2 used by SoapSampler
> >    - HC3CookieHandler
> >    - HttpHC3Impl
> >    - LoopbackHttpClientSocketFactory
> >    - SlowHttpClientSocketFactory
>
> OK
>
> >
> > --
> > Regards
> > Philippe
>