You are viewing a plain text version of this content. The canonical link for it is here.
Posted to dev@hc.apache.org by Oleg Kalnichevski <ol...@apache.org> on 2008/02/23 09:46:31 UTC

Test failures on Mac OS; was Re: [VOTE] HttpComponents Client 4.0-alpha3 release

Erik Abele wrote:
> On 20.02.2008, at 20:55, Oleg Kalnichevski wrote:
> 
>> Please vote on releasing these packages as HttpComponents Client
>> 4.0-alpha3. ...
>>
>> Packages:
>> http://people.apache.org/~olegk/httpclient-4.0-alpha/packages/
>>
>> ...
>> -------------------------------------------------------------------------- 
>>
>>  Vote:  HttpComponents Client 4.0-alpha3 release
>>  [X] +1 Release the packages as HttpComponents Client 4.0-alpha3.
>>  [ ] -1 I am against releasing the packages (must include a reason).
> 
> Some notes:
> 
> - Checksums and signatures are looking good
> 
> - Packages are matching the tag
> 
> - Packages have identical content except:
>   httpcomponents-client-4.0-alpha3-src.tar.gz
>   
> /module-client/src/main/java/org/apache/http/impl/conn/tsccm/doc-files/tsccm-structure.png 
> 
>   vs.
>   httpcomponents-client-4.0-alpha3-src.zip
>   
> /module-client/src/main/java/org/apache/http/impl/conn/tsccm/doc-files/tsccm-structure.png 
> 
> 
>   Differing on line endings (Unix/Windows); both PNGs are not usable!
>   Probably need to fiddle with src/main/assembly/*.xml to fix the assembly
>   process to not touch these binary files while adjusting line endings.
>   But certainly NOT a blocker!
> 

My bad. I messed up.


> - LICENSE/NOTICE/README/RELEASE_NOTES are looking good
> 
> - JAR contents are looking good
> 
> - Tests & build are fine on Linux but my OS X setup shows one
>   failure and several errors (though that's probably due to something 
> local)
> 

Here we seem to have a real issue. The test failures are easily 
reproducible on Mac OS. I just tested the latest HttpClient code on my 
wife's Mac Mini. All test cases that make use of the test HTTP server 
seem to be failing. I can look into problem no sooner than tomorrow in 
the eventing. I promised my wife to stay away from any kind of computer 
equipment for the best part of the weekend.

The bug seems to affect the test code only, so this still should not be 
a release blocker.

Oleg

java.net.SocketException: Socket closed
	at java.net.PlainSocketImpl.socketConnect(Native Method)
	at java.net.PlainSocketImpl.doConnect(PlainSocketImpl.java:333)
	at java.net.PlainSocketImpl.connectToAddress(PlainSocketImpl.java:195)
	at java.net.PlainSocketImpl.connect(PlainSocketImpl.java:182)
	at java.net.SocksSocketImpl.connect(SocksSocketImpl.java:432)
	at java.net.Socket.connect(Socket.java:520)
	at org.apache.http.conn.util.SocketUtils.connect(SocketUtils.java:104)
	at org.apache.http.conn.util.SocketUtils.connect(SocketUtils.java:70)
	at 
org.apache.http.conn.PlainSocketFactory.connectSocket(PlainSocketFactory.java:109)
	at 
org.apache.http.impl.conn.DefaultClientConnectionOperator.openConnection(DefaultClientConnectionOperator.java:134)
	at 
org.apache.http.impl.conn.AbstractPoolEntry.open(AbstractPoolEntry.java:149)
	at 
org.apache.http.impl.conn.AbstractPooledConnAdapter.open(AbstractPooledConnAdapter.java:119)
	at 
org.apache.http.impl.conn.ExecReqThread.doOpenConnection(ExecReqThread.java:132)
	at org.apache.http.impl.conn.ExecReqThread.run(ExecReqThread.java:102)
Tests run: 320, Failures: 1, Errors: 10, Skipped: 0, Time elapsed: 6.972 
sec <<< FAILURE!

Results :

Failed tests:
   testParallelRequests(org.apache.http.impl.conn.TestTSCCMWithServer)

Tests in error:
 
testRequestURIRewritingEmptyPath(org.apache.http.impl.client.TestRequestWrapper)
   testEcho(org.apache.http.impl.conn.TestLocalServer)
   testRandom(org.apache.http.impl.conn.TestLocalServer)
 
testCookieVersionSupportHeader1(org.apache.http.client.protocol.TestCookie2Support)
 
testSetCookieVersionMix(org.apache.http.client.protocol.TestCookie2Support)
   testBasicRedirect302(org.apache.http.client.protocol.TestRedirects)
   testBasicRedirect303(org.apache.http.client.protocol.TestRedirects)
   testBasicRedirect305(org.apache.http.client.protocol.TestRedirects)
   testRejectRelativeRedirect(org.apache.http.client.protocol.TestRedirects)
 
testRejectBogusRedirectLocation(org.apache.http.client.protocol.TestRedirects)

Tests run: 320, Failures: 1, Errors: 10, Skipped: 0





> Cheers,
> Erik
> 
> 
> ---------------------------------------------------------------------
> To unsubscribe, e-mail: dev-unsubscribe@hc.apache.org
> For additional commands, e-mail: dev-help@hc.apache.org
> 
> 


---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@hc.apache.org
For additional commands, e-mail: dev-help@hc.apache.org


Re: Test failures on Mac OS; was Re: [VOTE] HttpComponents Client 4.0-alpha3 release

Posted by Erik Abele <er...@codefaktor.de>.
On 23.02.2008, at 09:46, Oleg Kalnichevski wrote:

> Erik Abele wrote:
>> ...
>> - Packages have identical content except:
>>   httpcomponents-client-4.0-alpha3-src.tar.gz
>>   /module-client/src/main/java/org/apache/http/impl/conn/tsccm/doc- 
>> files/tsccm-structure.png   vs.
>>   httpcomponents-client-4.0-alpha3-src.zip
>>   /module-client/src/main/java/org/apache/http/impl/conn/tsccm/doc- 
>> files/tsccm-structure.png   Differing on line endings (Unix/ 
>> Windows); both PNGs are not usable!
>>   Probably need to fiddle with src/main/assembly/*.xml to fix the  
>> assembly
>>   process to not touch these binary files while adjusting line  
>> endings.
>>   But certainly NOT a blocker!
>
> My bad. I messed up.

Not a big issue, will have a look at fixing it next week if no one  
beats me to it.

>> - LICENSE/NOTICE/README/RELEASE_NOTES are looking good
>> - JAR contents are looking good
>> - Tests & build are fine on Linux but my OS X setup shows one
>>   failure and several errors (though that's probably due to  
>> something local)
>
> Here we seem to have a real issue. The test failures are easily  
> reproducible on Mac OS. I just tested the latest HttpClient code on  
> my wife's Mac Mini. All test cases that make use of the test HTTP  
> server seem to be failing. I can look into problem no sooner than  
> tomorrow in the eventing. I promised my wife to stay away from any  
> kind of computer equipment for the best part of the weekend.

Oh, interesting and I was expecting my local system to be at fault -  
thanks for looking into it!

> The bug seems to affect the test code only, so this still should  
> not be a release blocker.

Ok, I agree that it's not a release blocker, especially in the alpha  
phase :-)

Cheers,
Erik


---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@hc.apache.org
For additional commands, e-mail: dev-help@hc.apache.org


Re: Test failures on Mac OS; was Re: [VOTE] HttpComponents Client 4.0-alpha3 release

Posted by Erik Abele <er...@codefaktor.de>.
On 23.02.2008, at 10:17, Roland Weber wrote:

> Hi Oleg,
>
> Oleg Kalnichevski wrote:
>> Erik Abele wrote:
>>>
>>> - Tests & build are fine on Linux but my OS X setup shows one
>>>   failure and several errors (though that's probably due to  
>>> something local)
>> Here we seem to have a real issue. The test failures are easily  
>> reproducible on Mac OS. I just tested the latest HttpClient code  
>> on my wife's Mac Mini. All test cases that make use of the test  
>> HTTP server seem to be failing.
>
> I'm afraid that's my doing. I branched the LocalTestServer from
> core/TestHttpServer shortly before you added the expectation verifier
> there. I had a quick look just now and added several missing volatile
> declarations. Maybe I should stop writing Java MT code. Life was so
> much easier when I was still coding in assembler.
>
> I don't have a Mac to test with, so I can't help here. If it
> wasn't the volatiles,

No, it wasn't the volatiles - just svn up'ed and tested again -> same  
failure.

> the next candidates are timing issues in
> LocalTestServer.start/stop(). I've used the hardcoded IP address
> 127.0.0.1, but that should work on a Mac too.

Yes, it should ;-)

Cheers,
Erik


---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@hc.apache.org
For additional commands, e-mail: dev-help@hc.apache.org


Re: Test failures on Mac OS; was Re: [VOTE] HttpComponents Client 4.0-alpha3 release

Posted by Roland Weber <os...@dubioso.net>.
Hi Oleg,

> It turned out the problem had been introduced by this patch
> 
> http://svn.apache.org/viewvc?view=rev&revision=619163  
> 
> Please stop worrying about your ability to write MT code. It was all my
> fault.

not the missing volatiles in the test code...

cheers,
   Roland


---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@hc.apache.org
For additional commands, e-mail: dev-help@hc.apache.org


Re: Test failures on Mac OS; was Re: [VOTE] HttpComponents Client 4.0-alpha3 release

Posted by Sam Berlin <sb...@gmail.com>.
To spur some interest in the fisheye view (though it makes zero
difference), here's the comparable link:
http://fisheye6.cenqua.com/changelog/httpcomponents?cs=619163 .

Sam

On Mon, Feb 25, 2008 at 4:47 PM, Oleg Kalnichevski <ol...@apache.org> wrote:
>
> On Sat, 2008-02-23 at 10:17 +0100, Roland Weber wrote:
> > Hi Oleg,
> >
> > Oleg Kalnichevski wrote:
> > > Erik Abele wrote:
> > >>
> > >> - Tests & build are fine on Linux but my OS X setup shows one
> > >>   failure and several errors (though that's probably due to something
> > >> local)
> > >
> > > Here we seem to have a real issue. The test failures are easily
> > > reproducible on Mac OS. I just tested the latest HttpClient code on my
> > > wife's Mac Mini. All test cases that make use of the test HTTP server
> > > seem to be failing.
> >
> > I'm afraid that's my doing. I branched the LocalTestServer from
> > core/TestHttpServer shortly before you added the expectation verifier
> > there. I had a quick look just now and added several missing volatile
> > declarations. Maybe I should stop writing Java MT code. Life was so
> > much easier when I was still coding in assembler.
> >
>
> Roland,
>
> It turned out the problem had been introduced by this patch
>
> http://svn.apache.org/viewvc?view=rev&revision=619163
>
> Please stop worrying about your ability to write MT code. It was all my
> fault.
>
> Oleg
>
>
> > I don't have a Mac to test with, so I can't help here. If it
> > wasn't the volatiles, the next candidates are timing issues in
> > LocalTestServer.start/stop(). I've used the hardcoded IP address
> > 127.0.0.1, but that should work on a Mac too.
> >
> > cheers,
> >    Roland
> >
> >
> >
> > ---------------------------------------------------------------------
> > To unsubscribe, e-mail: dev-unsubscribe@hc.apache.org
> > For additional commands, e-mail: dev-help@hc.apache.org
> >
> >
>
>
> ---------------------------------------------------------------------
> To unsubscribe, e-mail: dev-unsubscribe@hc.apache.org
> For additional commands, e-mail: dev-help@hc.apache.org
>
>

---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@hc.apache.org
For additional commands, e-mail: dev-help@hc.apache.org


Re: Test failures on Mac OS; was Re: [VOTE] HttpComponents Client 4.0-alpha3 release

Posted by Oleg Kalnichevski <ol...@apache.org>.
On Mon, 2008-02-25 at 22:47 +0100, Oleg Kalnichevski wrote:
> On Sat, 2008-02-23 at 10:17 +0100, Roland Weber wrote:
> > Hi Oleg,
> > 
> > Oleg Kalnichevski wrote:
> > > Erik Abele wrote:
> > >>
> > >> - Tests & build are fine on Linux but my OS X setup shows one
> > >>   failure and several errors (though that's probably due to something 
> > >> local)
> > > 
> > > Here we seem to have a real issue. The test failures are easily 
> > > reproducible on Mac OS. I just tested the latest HttpClient code on my 
> > > wife's Mac Mini. All test cases that make use of the test HTTP server 
> > > seem to be failing.
> > 
> > I'm afraid that's my doing. I branched the LocalTestServer from
> > core/TestHttpServer shortly before you added the expectation verifier
> > there. I had a quick look just now and added several missing volatile
> > declarations. Maybe I should stop writing Java MT code. Life was so
> > much easier when I was still coding in assembler.
> > 
> 
> Roland,
> 
> It turned out the problem had been introduced by this patch
> 
> http://svn.apache.org/viewvc?view=rev&revision=619163  
> 
> Please stop worrying about your ability to write MT code. It was all my
> fault.
> 

I had to effectively revert HTTPCLIENT-643 changes in order to fix this
problem. Multihome failover on the socket factory level just does not
seem to work.

Oleg


> Oleg
> 
> > I don't have a Mac to test with, so I can't help here. If it
> > wasn't the volatiles, the next candidates are timing issues in
> > LocalTestServer.start/stop(). I've used the hardcoded IP address
> > 127.0.0.1, but that should work on a Mac too.
> > 
> > cheers,
> >    Roland
> > 
> > 
> > 
> > ---------------------------------------------------------------------
> > To unsubscribe, e-mail: dev-unsubscribe@hc.apache.org
> > For additional commands, e-mail: dev-help@hc.apache.org
> > 
> > 
> 
> 
> ---------------------------------------------------------------------
> To unsubscribe, e-mail: dev-unsubscribe@hc.apache.org
> For additional commands, e-mail: dev-help@hc.apache.org
> 
> 


---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@hc.apache.org
For additional commands, e-mail: dev-help@hc.apache.org


Re: Test failures on Mac OS; was Re: [VOTE] HttpComponents Client 4.0-alpha3 release

Posted by Oleg Kalnichevski <ol...@apache.org>.
On Sat, 2008-02-23 at 10:17 +0100, Roland Weber wrote:
> Hi Oleg,
> 
> Oleg Kalnichevski wrote:
> > Erik Abele wrote:
> >>
> >> - Tests & build are fine on Linux but my OS X setup shows one
> >>   failure and several errors (though that's probably due to something 
> >> local)
> > 
> > Here we seem to have a real issue. The test failures are easily 
> > reproducible on Mac OS. I just tested the latest HttpClient code on my 
> > wife's Mac Mini. All test cases that make use of the test HTTP server 
> > seem to be failing.
> 
> I'm afraid that's my doing. I branched the LocalTestServer from
> core/TestHttpServer shortly before you added the expectation verifier
> there. I had a quick look just now and added several missing volatile
> declarations. Maybe I should stop writing Java MT code. Life was so
> much easier when I was still coding in assembler.
> 

Roland,

It turned out the problem had been introduced by this patch

http://svn.apache.org/viewvc?view=rev&revision=619163  

Please stop worrying about your ability to write MT code. It was all my
fault.

Oleg

> I don't have a Mac to test with, so I can't help here. If it
> wasn't the volatiles, the next candidates are timing issues in
> LocalTestServer.start/stop(). I've used the hardcoded IP address
> 127.0.0.1, but that should work on a Mac too.
> 
> cheers,
>    Roland
> 
> 
> 
> ---------------------------------------------------------------------
> To unsubscribe, e-mail: dev-unsubscribe@hc.apache.org
> For additional commands, e-mail: dev-help@hc.apache.org
> 
> 


---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@hc.apache.org
For additional commands, e-mail: dev-help@hc.apache.org


Re: Test failures on Mac OS; was Re: [VOTE] HttpComponents Client 4.0-alpha3 release

Posted by Roland Weber <os...@dubioso.net>.
Hi Oleg,

Oleg Kalnichevski wrote:
> Erik Abele wrote:
>>
>> - Tests & build are fine on Linux but my OS X setup shows one
>>   failure and several errors (though that's probably due to something 
>> local)
> 
> Here we seem to have a real issue. The test failures are easily 
> reproducible on Mac OS. I just tested the latest HttpClient code on my 
> wife's Mac Mini. All test cases that make use of the test HTTP server 
> seem to be failing.

I'm afraid that's my doing. I branched the LocalTestServer from
core/TestHttpServer shortly before you added the expectation verifier
there. I had a quick look just now and added several missing volatile
declarations. Maybe I should stop writing Java MT code. Life was so
much easier when I was still coding in assembler.

I don't have a Mac to test with, so I can't help here. If it
wasn't the volatiles, the next candidates are timing issues in
LocalTestServer.start/stop(). I've used the hardcoded IP address
127.0.0.1, but that should work on a Mac too.

cheers,
   Roland



---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@hc.apache.org
For additional commands, e-mail: dev-help@hc.apache.org