headers
Johansson, Justin (Contractor | 1 Sep 2008 09:02
Picon

UIMACPP runtime problem on Linux

Hi Eddie & others,

I have previously run the UIMA CPE on Windows with a mix of Java and C++
annotators with success.
Now having built same annotators for Linux, the UIMA CPE spits when
encountering the first C++
Annotator with message like:

'No uima on java.library.path'

And a stack track indicating exceptions thrown by load (native) library
in java.

I'm guessing this means uima cannot find libuima.so or such but that is
there in the same directory as the annotator .so

Interestingly runAECpp MyCppAnnotatorDescriptor.xml works .. the problem
is just when trying to run the whole show.

Seems like I must be missing something in the UIMA C++ setup for Unix
(Linux).

Any ideas?

Thanks in advance,

Justin Johansson

IMPORTANT: This email remains the property of the Australian Defence Organisation and is subject to the
jurisdiction of section 70 of the CRIMES ACT 1914.  If you have received this email in error, you are
(Continue reading)

Permalink | Reply | 3 comments |
headers
Ashutosh Sharma | 2 Sep 2008 06:11
Picon
Favicon

Multi-threaded CAS Consumer


Hi,

I would like to run CPE in multithreaded environment. For this purpose, I changed the value of the variable
named processingUnitThreadCount=3 in CPE Descriptor file. When I run the CPE, it gave me a Java null
pointer exception in my CAS Consumer. Where I initialize all basics open connection and initialize the
command to insert the data. 
Please help me to overcome this issue. How can I achieve the running CPE in multithreaded mode. 

Thanks & Regards,
Ashu

_________________________________________________________________
Movies, sports & news! Get your daily entertainment fix, only on live.com
http://www.live.com/?scope=video&form=MICOAL
Permalink | Reply | 4 comments |
headers
Burn Lewis | 2 Sep 2008 14:17
Picon

Re: UIMACPP runtime problem on Linux

Justin,

I think this means that libuima.so is not in the LD_LIBRARY_PATH ... do you
have $UIMACPP_HOME/lib there?
If this is not the problem, have you been able to run the examples as
described in the README?

Burn.

On Mon, Sep 1, 2008 at 3:02 AM, Johansson, Justin (Contractor) <
Justin.Johansson@...> wrote:

> Hi Eddie & others,
>
> I have previously run the UIMA CPE on Windows with a mix of Java and C++
> annotators with success.
> Now having built same annotators for Linux, the UIMA CPE spits when
> encountering the first C++
> Annotator with message like:
>
> 'No uima on java.library.path'
>
> And a stack track indicating exceptions thrown by load (native) library
> in java.
>
> I'm guessing this means uima cannot find libuima.so or such but that is
> there in the same directory as the annotator .so
>
> Interestingly runAECpp MyCppAnnotatorDescriptor.xml works .. the problem
> is just when trying to run the whole show.
(Continue reading)

Permalink | Reply | 2 comments |
headers
Burn Lewis | 2 Sep 2008 14:40
Picon

Re: Multi-threaded CAS Consumer

Does your CAS Consumer have the default of false for
multipleDeploymentAllowed?  Usually only the annotator pipeline is run in
multiple threads ... the CPE runs a single-threaded Collection Reader that
feeds the multiple replicated pipelines which all feed into a
single-threaded CAS Consumer.  I think you can run multiple CAS Consumers
but then you are resposible for serializing access to your shared output
resources.

Perhaps you could post the full trace of your exception.

Burn.

On Tue, Sep 2, 2008 at 12:11 AM, Ashutosh Sharma <ashu.sharma@...>wrote:

>
> Hi,
>
> I would like to run CPE in multithreaded environment. For this purpose, I
> changed the value of the variable named processingUnitThreadCount=3 in CPE
> Descriptor file. When I run the CPE, it gave me a Java null pointer
> exception in my CAS Consumer. Where I initialize all basics open connection
> and initialize the command to insert the data.
> Please help me to overcome this issue. How can I achieve the running CPE in
> multithreaded mode.
>
> Thanks & Regards,
> Ashu
>
>
> _________________________________________________________________
(Continue reading)

Permalink | Reply | 3 comments |
headers
Ashutosh Sharma | 2 Sep 2008 14:58
Picon
Favicon

RE: Multi-threaded CAS Consumer


Thanks Burn,

Yes, the default was true for multipleDeploymentAllowed. Thanks for your response. 
Now I am able to run CPE successfully. Thanks a lot. 
Wait a minute, You have said that we can run multiple CAS Consumers but need serialization to access the
shared output. 
Could you please tell me, how can I achieve this. Do you suggest some reading for this issue.

Thanks & Regards,
Ashu

> Date: Tue, 2 Sep 2008 08:40:01 -0400
> From: burnlewis@...
> To: uima-user@...
> Subject: Re: Multi-threaded CAS Consumer
> 
> Does your CAS Consumer have the default of false for
> multipleDeploymentAllowed?  Usually only the annotator pipeline is run in
> multiple threads ... the CPE runs a single-threaded Collection Reader that
> feeds the multiple replicated pipelines which all feed into a
> single-threaded CAS Consumer.  I think you can run multiple CAS Consumers
> but then you are resposible for serializing access to your shared output
> resources.
> 
> Perhaps you could post the full trace of your exception.
> 
> Burn.
> 
> On Tue, Sep 2, 2008 at 12:11 AM, Ashutosh Sharma <ashu.sharma <at> hotmail.com>wrote:
(Continue reading)

Permalink | Reply | 2 comments |
headers
Burn Lewis | 2 Sep 2008 15:33
Picon

Re: Multi-threaded CAS Consumer

Ashu,

Glad to hear it's working now.  I don't have much experience but I've found
"Java Concurrency in Practice" by Brian Goetz very helpful and thorough.

Regards,  Burn.

On Tue, Sep 2, 2008 at 8:58 AM, Ashutosh Sharma <ashu.sharma@...>wrote:

>
> Thanks Burn,
>
> Yes, the default was true for multipleDeploymentAllowed. Thanks for your
> response.
> Now I am able to run CPE successfully. Thanks a lot.
> Wait a minute, You have said that we can run multiple CAS Consumers but
> need serialization to access the shared output.
> Could you please tell me, how can I achieve this. Do you suggest some
> reading for this issue.
>
> Thanks & Regards,
> Ashu
>
>
>
> > Date: Tue, 2 Sep 2008 08:40:01 -0400
> > From: burnlewis@...
> > To: uima-user@...
> > Subject: Re: Multi-threaded CAS Consumer
> >
(Continue reading)

Permalink | Reply | 1 comment |
headers
Ashutosh Sharma | 2 Sep 2008 15:44
Picon
Favicon

RE: Multi-threaded CAS Consumer


Thanks Burn

Regards,
Ashu

> Date: Tue, 2 Sep 2008 09:33:22 -0400
> From: burnlewis@...
> To: uima-user@...
> Subject: Re: Multi-threaded CAS Consumer
> 
> Ashu,
> 
> Glad to hear it's working now.  I don't have much experience but I've found
> "Java Concurrency in Practice" by Brian Goetz very helpful and thorough.
> 
> Regards,  Burn.
> 
> On Tue, Sep 2, 2008 at 8:58 AM, Ashutosh Sharma <ashu.sharma <at> hotmail.com>wrote:
> 
> >
> > Thanks Burn,
> >
> > Yes, the default was true for multipleDeploymentAllowed. Thanks for your
> > response.
> > Now I am able to run CPE successfully. Thanks a lot.
> > Wait a minute, You have said that we can run multiple CAS Consumers but
> > need serialization to access the shared output.
> > Could you please tell me, how can I achieve this. Do you suggest some
> > reading for this issue.
(Continue reading)

Permalink | Reply | 0 comments |
headers
Johansson, Justin (Contractor | 3 Sep 2008 04:51
Picon

RE: UIMACPP runtime problem on Linux


Thanks for that Burn.  That put me on the right track.  Turns out that
the libuima.so was a link to a non-existent target so technically
speaking you were correct and libuima.so was not in the LD_LIBRARY_PATH.

Having solved that one, the C++ annotator is now trying to load but we
spit a different error message now about wrong ELFCLASS64 in respect of
libuima.so.

The box this is running on has an Athlon 64 bit processor but afaik the
OS and installed software is all 32 bit.

Sounds now like we need to explicitly build libuima.so as 32 bit.
What/where is the configuration switch for that.

Many thanks for your help.

Justin Johansson

-----Original Message-----
From: Burn Lewis [mailto:burnlewis@...] 
Sent: Tuesday, 2 September 2008 21:47
To: uima-user@...
Subject: Re: UIMACPP runtime problem on Linux

Justin,

I think this means that libuima.so is not in the LD_LIBRARY_PATH ... do
you have $UIMACPP_HOME/lib there?
If this is not the problem, have you been able to run the examples as
(Continue reading)

Permalink | Reply | 1 comment |
headers
Burn Lewis | 3 Sep 2008 15:46
Picon

Re: UIMACPP runtime problem on Linux

To build a 32-bit annotator on a 64-bit OS you can uncomment the FORCE_32
entry in base.mak.

But when building libuima.so the configure & automake steps should have
determined what your OS is and set the appropriate flags.  You could try
running file & ldd on the uima library and your annotator to check that
they're compatible and can find all their dependencies.

I think one of our test machines is 64-bit with a 32-bit OS.

Burn.

On Tue, Sep 2, 2008 at 10:51 PM, Johansson, Justin (Contractor) <
Justin.Johansson@...> wrote:

>
> Thanks for that Burn.  That put me on the right track.  Turns out that
> the libuima.so was a link to a non-existent target so technically
> speaking you were correct and libuima.so was not in the LD_LIBRARY_PATH.
>
> Having solved that one, the C++ annotator is now trying to load but we
> spit a different error message now about wrong ELFCLASS64 in respect of
> libuima.so.
>
> The box this is running on has an Athlon 64 bit processor but afaik the
> OS and installed software is all 32 bit.
>
> Sounds now like we need to explicitly build libuima.so as 32 bit.
> What/where is the configuration switch for that.
>
(Continue reading)

Permalink | Reply | 0 comments |
headers
Peter Klügl | 4 Sep 2008 13:25
Picon
Favicon

Re: annotation comparator

Hi,

what is the status quo for the comparison of two CAS right now? Is there 
yet any usable solution (with or without documentation)?

I am developing a rule-based system (with scripting functionalities) 
especially for complex information and text extraction tasks. The IDE is 
DLTK-based and UIMA descriptors (for a generic implementation) are 
generated automatically. Currently i am improving a information 
extraction application with a test-driven approach. The test cases are, 
of course, CAS XMI files and the comparison (of two CAS) is working, but 
yet unsatisfying. I am especially interested in annotations for the 
false positives and false negatives (overlapping or not overlapping).

Back to my question:
How do you all compare two CAS?
Is there a reusable implementation?

Peter

Katrin Tomanek schrieb:
> Hi,
>
>
>> Depends what your favorite tooling story is.  If you prefer
>> the eclipse tooling, it should go into eclipse.  I know
>> people who would use this kind of functionality if it was
>> in CVD :-)
>>
>>>
(Continue reading)

Permalink | Reply | 4 comments |

Gmane