[kepler-dev] GARP actor issues

Dan Higgins higgins at nceas.ucsb.edu
Tue Jul 28 19:19:06 PDT 2009


Tim,
    I originally did much of the GARP work in Kepler (but I haven't been 
involved noin Kepler at all  for 18 months or so). The C++ code was 
based on the PC  GARP code developed at University of Kansas. I think 
Chad Berkley worked on converting it to run under JNI (a long time ago). 
My memory is that the code was always a bit fragile and dependencies on 
different platforms were always a problem.

Dan Higgins


Timothy McPhillips wrote:
> All,
>
> Is anyone else using the GARP actors, or in a position to help debug 
> the problem with the GARPPresampleLayers actor described below?  I 
> have reproduced the problem at the trunk, and stepped through the Java 
> portion of the code in a debugger, but it appears that the issue may 
> be in the C++ code the actor employs via JNI.  Is anyone already set 
> up to build this code and its dependencies on our various supported 
> platforms?
>
> Thanks very much!
>
> Tim
>
> On Jul 27, 2009, at 11:24 AM, Timothy McPhillips wrote:
>
>> Hi Russell,
>>
>> Ok, I think I've reproduced your problem with the GARPPresampleLayers 
>> actor in Windows XP SP3, Vista SP1 (32-bit),  and MacOS 10.5.7.  In 
>> each case specifying a file for layersetFilenameParameter that is not 
>> in lib/testdata/garp causes Kepler to crash when the workflow 
>> containing the actor is run.  Specifying a file that does not exist 
>> at all gives the same error. Specifying files renamed within the 
>> lib/testdata/garp directory seems to be fine, as you say.
>>
>> On Ubuntu 9.0.4, things do not get this far.  It looks like the 
>> system cannot find a required shared library. (libexpat.so ?)
>>
>> I'm going to see if I can reproduce this in my development 
>> environments so I can track down the problem.
>>
>> Cheers,
>>
>> Tim
>>
>> On Jul 26, 2009, at 1:58 PM, R.P.Mciver at cs.cardiff.ac.uk wrote:
>>
>>> Hi Tim,
>>>
>>> Thanks for the response. Hopefully the following should be enough 
>>> information to recreate the situation:
>>>
>>> The OS is Windows XP (SP3 if we're being picky!)
>>>
>>> Kepler is installed at C:\Program Files\Kepler-1.0.0 (what I believe 
>>> to be the default for the installer)
>>>
>>> Workflow created consists of simply an SDF Director set to 1 
>>> iteration and the actor causing the issue, GARPPresampleLayers, with 
>>> the following Parameters:
>>>
>>> * layerSetFilenameParameter: "C:\Program 
>>> Files\Kepler-1.0.0\lib\testdata\garp\IPCC.dxl"
>>>
>>> (this is the one that seems to cause the error, functions fine when 
>>> set to this file, or to one of the similar files in the same 
>>> location, or if set to the same file renamed in that location. Seem 
>>> to fail if set to anything in another location - even if it's the 
>>> same file. Also fails if set to the same file in a subdirectory of 
>>> this location)
>>>
>>> * dataPointFileNameParameter: "C:\Program 
>>> Files\Kepler-1.0.0\lib\testdata\garp\digir_data_mephitis"
>>>
>>> * cellSetFileNameParameter: C:\Program Files\Kepler-1.0.0\CellSet.xml
>>>
>>>
>>> These input data files were just taken from examining the setup of 
>>> the same actor in the "$Kepler\lib\demos\ENM" directory. Also 
>>> checked if the problem was limited to the actors paramater, but the 
>>> same result occurs if trying to provide the input file using the 
>>> corresponding input port as well.
>>>
>>> Hope this is enough information. Thanks again for the help working 
>>> this one out!
>>>
>>> Russell
>>>
>>>
>>> Quoting Timothy McPhillips <tmcphillips at mac.com>:
>>>
>>>> Hi Russell,
>>>>
>>>> I've never used the GARP actors or workflows, but I'd be happy to 
>>>> try to reproduce the problem you're seeing and help track down the 
>>>> problem (assuming no one else knows the answer to your question 
>>>> already).
>>>>
>>>> Can you send me enough information to reproduce your environment 
>>>> (OS, location of Kepler installation, etc), along with the input 
>>>> data file?
>>>>
>>>> Thanks!
>>>>
>>>> Tim
>>>>
>>>> On Jul 26, 2009, at 10:56 AM, R.P.Mciver at cs.cardiff.ac.uk wrote:
>>>>
>>>>> Hi all,
>>>>>
>>>>> Hopefully someone can help me out here, I've playing about with 
>>>>> the kepler UI (mainly just looking at how things work etc.) and 
>>>>> thought I'd try using my slightly modified GUI to compose some 
>>>>> workflows utilisng the GARP set of actors, similar to the niche 
>>>>> modelling examples that come with kepler.
>>>>>
>>>>> Unfortunately I've been getting some errors (well, stack overflows 
>>>>> to be precise). After spending a long time thinking it was as a 
>>>>> result of something I'd changed in the UI, I went back and tried 
>>>>> the same on a clean install of the 1.0.0 release, only to find 
>>>>> there were still some problems.
>>>>>
>>>>> It seems that whenever I provide the GARPPresampleLayers actor 
>>>>> with a "layerset" input file which is located anywhere other than 
>>>>> the "$Kepler/lib/testdata/garp" folder then kepler crashes out 
>>>>> with a stack overflow.
>>>>>
>>>>> The only difference in the error between using "vanilla" kepler 
>>>>> 1.0.0 and my modified version is that when attempting to execute 
>>>>> the actor in my code I get the stack overflow regardless of where 
>>>>> the layerset file is located.
>>>>>
>>>>> Any ideas what is going on here - is the location for where the 
>>>>> file has to be defined somewhere? I've tried just executing the 
>>>>> GARPPresampleLayers actor in isolation so I'm fairly certain it's 
>>>>> not a problem with some other part of a workflow.
>>>>>
>>>>> Any help would be appreciated!
>>>>>
>>>>> Cheers,
>>>>>
>>>>> Russell
>>>>>
>>>>> =======================================================
>>>>> Russell McIver
>>>>> Research Student
>>>>> Room C/2.14
>>>>> Cardiff School of Computer Science, Cardiff University,
>>>>> 5 The Parade, Roath, Cardiff, CF24 3AA, UK.
>>>>> =======================================================
>
> _______________________________________________
> Kepler-dev mailing list
> Kepler-dev at kepler-project.org
> http://mercury.nceas.ucsb.edu/kepler/mailman/listinfo/kepler-dev
>



More information about the Kepler-dev mailing list