hc-httpclient-users mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Alexander Bernstein <alexbernstein1...@yahoo.com.INVALID>
Subject Re: GPF in WindowsNegotiateScheme.dispose()
Date Fri, 28 Aug 2015 19:22:41 GMT
Thank you, Michael.
I too noticed that the code was marked as experimental. I have a workaround for this issue
(simply keep the references to the created WindowsNegotiateScheme objects and dispose of them
before GC gets to them. This works for me.
Having said that, should I expect more problems later? If these are known issues, are there

Now, this code is experimental which I can understand, so I ran an experiment and found a
problem. But putting something untested/not reviewed into production (the released libraries
in this case) seems dangerous to me. I am not questioning your practices or anything, just
a side comment.
      From: Michael Osipov <michaelo@apache.org>
 To: HttpClient User Discussion <httpclient-users@hc.apache.org> 
 Sent: Friday, August 28, 2015 3:04 PM
 Subject: Re: GPF in WindowsNegotiateScheme.dispose()
Am 2015-08-26 um 22:07 schrieb Alexander Bernstein:

> Hello,
> I am new here, please pardon if this is not the right place for this kind of question.
> I am using httpclient-win-4.5 from httpClient 4.5 and jna-4.1.0/jna-platform-4.1.0 JNA
libraries to authenticate to Kerberos server from Eclipse-based application.
> I create my client from WinHttpClients.custom(). The authentication is successful, but
sooner or later, when JVM calls WindowsNegotiateScheme.dispose() my Eclipse crashes with GPF.
> I noticed that WindowsNegotiateSchemeFactory.create() is called twice, for some reason.
Not sure if this is a normal flow and/or is relevant to the problem.

I hav obstained this discussion but I think now it's time to clarify 
some issues:

1. The module is marked as experimental
2. The code provided (SSPI-based auth) has been simply copied over from 
JNA's examples *without* being reviewed.
3. The code as well as JNA contain bugs which do not make this scheme 
suitable for production.
4. HttpClient's current auth interfaces do not allow to implement that 

Most of those issues I have layed out on the dev mailing list already. 
Please search for.


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

  • Unnamed multipart/alternative (inline, None, 0 bytes)
View raw message