ode-user mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Ciaran <ciar...@gmail.com>
Subject Re: InstanceManagement.ListInstances, filter broken?
Date Wed, 21 May 2008 15:58:58 GMT
On Wed, May 21, 2008 at 3:54 PM, Matthieu Riou <matthieu@offthelip.org>
wrote:

> On Wed, May 21, 2008 at 12:15 AM, Ciaran <ciaranj@gmail.com> wrote:
>
> > Sadly it doesn't complete for me :( If I move the 'State-Simple_Inbound'
> > sequence into the if clause of the 'TRansitionResolver2' element, and
> > remove
> > the 'transition4' link (which is no longer needed) then it does complete,
> > very strange :(
> >
>
> I started having a look yesterday but was doing too many things in parallel
> and haven't had time to identify the problem yet. But tomorrow is another
> day :)
>
>
> >
> > Is there an alternative construct to flow which allows me to route
> between
> > different activities, I don't care about concurrency none of my
> activities
> > need to be executed 'concurrently', sequence doesn't let me move about
> > between activities, I've tried scope but when using a scope none of the
> > contained activities can see the links defined in the parent
> flow/sequence
> > etc.   Is scope what I need, and should I just try and persevere with it
> ?
> >
>
> What about writing what you want to do in pseudo-code so we can understand
> what you want to do? From there we can probably suggest a way to model your
> BPEL.

Ironically there is no pseudo-code :(   I'm converting a State-transition
network to BPEL, where each state is a sequence of steps.  Each state can
lead to one or more other states until a terminal state is reached. (There
is some logic in there to remove cyclic paths, so effectively its an
acylic-graph)

A state can only execute one at a time, so there is no concurrency, i.e I
would like to model:

                       ,D
                     .'
A -> B ->C  '--------F
                     `\
                        \,
                         E
Where the route to D  E or F can be conditional,   currently A,B,C, D, E and
F are represented as indvidual 'sequences' inside a flow with the
transitions beween them done as links.  If there are conditional links
leading from a state  then that state will have a link to an 'if' element
which contains the relevant tests and the source element of a link to the
'next' correct state.

Does this make sense at all ?
Many Thanks
- Cj.


>
>
> Cheers,
> Matthieu
>
>
> >
> > Many thanks
> > - Ciaran.
> >
> >
> > On Tue, May 20, 2008 at 11:45 PM, Alex Boisvert <boisvert@intalio.com>
> > wrote:
> >
> > > I just took a look at the BPEL and it looks valid.  The flow should
> > > complete
> > > after all the links have fired.  We'll need to run it and see what
> > actually
> > > happens to determine the issue.
> > >
> > > alex
> > >
> > > On Tue, May 20, 2008 at 12:11 PM, Ciaran <ciaranj@gmail.com> wrote:
> > >
> > > > As promised the attached file (really a .zip file, stupid virus
> > scanners)
> > > > contains a ready-to-deploy testcase that will never return :( .. What
> > am
> > > I
> > > > doing wrong!??! (Please  note the BPEL *is* a bit wrong as I've just
> > > hacked
> > > > it together to demonstrate my issue (as in I'm not assign the
> > > receive/reply
> > > > variables correctly!) but that shouldn't be relevant.
> > > >
> > > > Any clues? Please, if I can't get this working I have to abandon my
> > > > investigations with ODE which is a shame as I was pretty keen until
> > now!
> > > > (Hopefully its just my rubbish BPEL skills!)
> > > > -Ciaran
> > > >
> > > >
> > > >
> > > >
> > > >
> > > > On Tue, May 20, 2008 at 7:09 PM, Alex Boisvert <boisvert@intalio.com
> >
> > > > wrote:
> > > >
> > > >> On Tue, May 20, 2008 at 11:01 AM, Ciaran <ciaranj@gmail.com>
wrote:
> > > >>
> > > >> > Hi,
> > > >> > Thanks for looking into this Alex, can I configure ODE to use
the
> > > >> Hibernate
> > > >> > DAO for the Derby DB it ships with, or am I left using the OpenJPA
> > > >> version?
> > > >>
> > > >>
> > > >> You can use the Hibernate DAO by setting,
> > > >>
> > > >>
> > > >>
> > >
> >
> ode-axis2.dao.factory=org.apache.ode.daohib.bpel.BpelDAOConnectionFactoryImpl
> > > >>
> > > >> in your ode-axis2.properties file.  (And using the Hibernate
> database
> > > >> schemas)
> > > >>
> > > >>
> > > >> > Also I'm having some serious (blocking :( ) issuess getting my
> BPEL
> > > >> working
> > > >> > in ODE, I *think* I've finally tracked the issue down to a flow
> that
> > > has
> > > >> > several complex routes through it (links marshalled by 'if'
> > elements),
> > > >> it
> > > >> > seems that if any of the top-level elements in the flow element
> are
> > > >> targets
> > > >> > of links who's sources are 'empty' elements in some of my elseif
> > > >> elements
> > > >> > elsewhere, and these elseif elements do not execute then my BPEL
> > flow
> > > >> sits
> > > >> > waiting for these links to 'falsify'  (which will never happen
as
> > the
> > > >> > workflow has effectively finished going down the expected route).
> > > >> >
> > > >> > My understand of these things isn't great, but I think the issue
> is
> > > >> whether
> > > >> > DPE is being successfully applied in this particular situation
(I
> > can
> > > >> find
> > > >> > a
> > > >> > DPE issue filed in JIRA for when sequences fault, but I don't
> think
> > > this
> > > >> is
> > > >> > my case exactly).   I'm trying to create the smallest test case
so
> > > with
> > > >> > luck
> > > >> > someone will be able to tell me that what i'm trying to do in
BPEL
> > is
> > > >> > illegal (it should be noted however that it works in other BPEL
> > > engines
> > > >> :(
> > > >> > )
> > > >>
> > > >>
> > > >> Yep, a small reproducible test case is key here...
> > > >>
> > > >> alex
> > > >>
> > > >
> > > >
> > > >
> > > > --
> > > > - Ciaran
> > >
> >
> >
> >
> > --
> > - Ciaran
> >
>



-- 
- Ciaran

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