[kepler-dev] java cog jar files duplication .....

[jagan]

Hi Dan,

It won't interfere or pollute with kepler's classpath only thing is 
there would be number duplication of jar files in cvs repository.
I am planning to put cog and its dependent third party jar files in 
$KEPLER/cog directory, unless we specify explicitly these
jars won't appear in kepler classpath.

with regards,

Jagan Kommineni


Dan Higgins wrote:

> jagan,
>     From your example code, it appears that you are launching another 
> JVM with its own classpath. That operation should be independent of 
> any of Kepler's parameters, including the Kepler classpath. So I don't 
> see how this secondary Java process can be interferred with by Kepler ???
>
> Dan Higgins
>
> jagan wrote:
>
>> Hi Ilkay,
>>
>>
>> Ilkay Altintas wrote:
>>
>>>On Jun 22, 2006, at 12:33 AM, jagan wrote:
>>>
>>>  
>>>
>>>>Hi Ilkay,
>>>>
>>>>I am getting some duplicated jar files with the latest version of  
>>>>java cog toolkit cog-4_1_4.
>>>>As you know, cog supports number of different versions of globus  
>>>>toolkits. They are loading some java classes
>>>>based on the relative path of  the other jar files (those jars are  
>>>>not in the classpath).  That means
>>>>    
>>>>
>>>
>>>Jagan,
>>>
>>>Was this not happening before? What initiated the problem?
>>>
>>>  
>>>
>> I think this problem only with the latest version, it is not the case 
>> before, you may have look at the cog's documentation.
>>
>>http://wiki.cogkit.org/index.php/FAQ#How_do_I_integrate_the_CoG_Jars_into_my_own_project.3F
>>
>>
>>>>I have to maintain their directory structure as tease  at least for  
>>>>some  jar files.
>>>>    
>>>>
>>>
>>>I couldn't understand this. Could you explain it further?
>>>
>>>  
>>>
>> Cog's 4_1_4 library structure is some thing like this
>> jagan at brama ~/cogkit/cog/dist/cog-4_1_4
>> $ ls
>> bin  etc  examples  lib  lib-gt3_2_1  lib-gt4_0_0
>> There are number of jars in lib folder which are required to be in 
>> classpath and there other jar files in folders lib-gt3_2_1 and  
>> lib-gt4_0_0
>> These folders are required to maintain the relative path but those 
>> shouldn't go in "classpath". 
>>
>>
>>>>I am  using their services  by calling an external process from the  
>>>>kepler actor.
>>>>I can create a completely seperate classpath based on cog jar files.
>>>>That means I don't need pollute Kepler classpath from the cog  
>>>>classpath.
>>>>    
>>>>
>>>
>>>You mean you won't include them in the classpath? Maybe you create a  
>>>new configuration for your project?
>>>
>>>  
>>>
>> I would like to put all the cog jars in $KEPLER/cog folder which is 
>> not a regular kepler class path folder.
>> I create a new classpath for cog taks from with in the kepler actor 
>> and call a new sub process to handle this task as shown below.
>>
>> String wfCommnad =  "java -classpath "+cogClassPath+" 
>> org.globus.cog.karajan.Loader "+ xmlScript;
>> process = Runtime.getRuntime().exec(wfCommnad);
>>
>>>>Already I am storing cog's jar files in $KEPLER/cog directory by  
>>>>using third party jars from kepler classpath.
>>>>With the latest version I need to maintain relative path as I  
>>>>mention before which ends up with some duplication in jar files of  
>>>>Kepler's cvs repository.
>>>>    
>>>>
>>>
>>>I don't think relative paths is a good idea. I think it is better to  
>>>go with a separate configuration file. That is if I understood you  
>>>correctly.
>>>
>>>  
>>>
>> I agree with you the relative path is not a good solution.
>>
>> with regards,
>>
>> Jagan Kommineni
>>
>>>HTH,
>>>-ilkay
>>>
>>>  
>>>
>>>>Could you mind to give some advise please?
>>>>
>>>>with regards,
>>>>
>>>>Jagan Kommineni
>>>>
>>>>    
>>>>
>>>
>>>
>>>  
>>>
>>
>>------------------------------------------------------------------------
>>
>>_______________________________________________
>>Kepler-dev mailing list
>>Kepler-dev at ecoinformatics.org
>>http://mercury.nceas.ucsb.edu/ecoinformatics/mailman/listinfo/kepler-dev
>>  
>>
>
>------------------------------------------------------------------------
>
>_______________________________________________
>Kepler-dev mailing list
>Kepler-dev at ecoinformatics.org
>http://mercury.nceas.ucsb.edu/ecoinformatics/mailman/listinfo/kepler-dev
>  
>

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mercury.nceas.ucsb.edu/kepler/pipermail/kepler-dev/attachments/20060623/7b258889/attachment.html>