Re: Some questions

Hello all

David wrote:
> Hi Geeta.
> 
> first, thanks for information.
> 
> I'd like a bit of comment around the multiple ways to access the
> engine. Isn't it possible to tell in configuration of engine
> something like 'use wrapper call which will act as a proxy to a
> remote corba engine' ? I thought it would be possible due to the 
> modularity of shark engine (something like a  'corba client engine').

No. Shark isn't *that* modular, if client code uses POJO interfaces it
uses them, same applies for CORBA. Although we thought about this (we 
would have one instead of two admin applications) we never got to make 
such feature.

> Having multiple engine with different name woudln't mess up think
> when deploying new xpdl (eg, one engine is using the old xpdl while
> the admin engine will get the new one, resulting in desynchronization
> between engines?)

See PackageAdministration.synchronizeXPDLCache()

> Moreover, i would like my JSP to use the external engine (thru corba
> for example) to reduce the memory usage in tomcat. One of our web
> applications is quite unstable and sometimes eats all memory, i
> wouldn't like the heavy processing in workflow to be done inside
> tomcat.

Ok. That's your choice.

> On point number 5, you send me to initialising database, there is
> some shell script to run? Isn't there a simple SQL script i could
> send to database to create table/sequences/triggers? According to
> logs i saw when running the script it also add demo workflow instance
> iin database. Nice for tests, less for production...

I hope it doesn't upload anything into database. Scripts bin/recreateDB
and configure should only drop and create data model on chosen JDBC
connection - nothing more, nothing less. Anyway sql scripts are in
Shark/conf/sql/DBVENDOR/sql directory.

> Last comment, do you mean by your suggested code that, if i don't
> call it, all workflows started on the old version of the xpdl will be
> broken until i upgrade them using code? If yes, shouldn't it be done
> automatically by engine when new xpdl is uploaded?

All instances started with old xpdl version will continue executing
according to that xpdl. All new instances *may* use whatever xpdl
version your want.

> Le Mardi 19 Juillet 2005 15:38, [email protected] a écrit :
> 
>> Hi David,
>> 
>> Here are answers to some of your questions:
>> 
>> 2. http://shark.objectweb.org/doc/1.1/ToolAgents/tool_agents.html 
>> 3. I don't think this is possible.. at least as far as i know. 4.
>> The key is to use different Shark.confs with different engine names
>> and to turn all caches off: 
>> http://mail-archive.objectweb.org/shark/2005-02/msg00071.html 5.
>> http://shark.objectweb.org/doc/1.1/HowTo/how_to.html#d0e11 6. You
>> can update the current version of a process while old processes are
>>  still running against an older version. Code you can use:
>> 
>> PackageAdministration pa = 
>> Shark.getInstance().getAdminInterface().getPackageAdministration();
>>  RepositoryMgr rm = Shark.getInstance().getRepositoryManager();
>> 
>> String pkgId = rm.getPackageId(xpdlName); try { 
>> pa.updatePackage(pkgId, xpdlName); } catch (Throwable e) { throw
>> new BaseException(e); }
>> 
>> Regards, Geeta
>> 
>> 
>> delbd <[email protected]> 07/19/2005 04:45 AM Please respond to 
>> [email protected]
>> To [email protected] cc
>> 
>> Subject {Spam?} [shark] Some questions
>> 
>> Hello, i'd like to get the following informations.
>> 
>> 1) How to have shark use the users/roles i have define in a HP-UX
>> system. I already have code which is able to retrieve such
>> users/group in a tomcat realm, i need to know what i have to
>> implement so shark recognoize those users/group as participants

Shark modularity kicks in here ;-)
You should make new implementations of  Shark's UserGroupManager and
AuthenticationManager interfaces and change Shark.conf to refer to new
class names. Shark already has two implementations: DODS and LDAP.

>> 2) How to add custom tools? (like pdf generation)
>> 
>> 3) Once this is done, is it possible to have the current list of 
>> participants/tools preloaded in jawe when a wf admin wants to
>> create a new package?
>> 
>> 4) what is the suggested configuration if i want the shark engine
>> to run in it's own system process while I have a separate JSP based
>> application access this engine (through corba if am not wrong)
>> 
>> 5) Is it easy to use something else than Hypersonic SQL as database
>> (are there some SQL script somewhere to create the database?)
>> 
>> 6) Is shark quite resistant to changes in process definition while
>> some workflow instance are running on old versions of those
>> process?
>> 
>> Thanks -- David Delbecq Royal Meteorological Institute of Belgium
>> 
>> - Is there life after /sbin/halt -p?

hope this helps
-- 
Vlada

--
You receive this message as a subscriber of the [email protected] mailing list.
To unsubscribe: mailto:[email protected]
For general help: mailto:[email protected]?subject=help
ObjectWeb mailing lists service home page: http://www.objectweb.org/wws