commons-user mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Roger Hoover" <roger.hoo...@gmail.com>
Subject Re: SCXML Templating
Date Thu, 08 Jan 2009 20:02:58 GMT
Hi Rahul,

Please enlighten me if I'm missing something.

It seems like the only reason that state names must be globally unique is
that they need to be referenced by name in transitions.  Why not reference
them by something similar to XPath?  In the example you gave earlier, the
states could be uniquely identified by expressions such as "s1/foo, s1/bar,
and s2/foo, s2/bar?

In this way, the names would need to be unique only in their scope but not
unique globally.  I think it would make modularizing SCXML easier.  What you
think?

Roger


On Thu, Jan 8, 2009 at 11:22 AM, Rahul Akolkar <rahul.akolkar@gmail.com>wrote:

> On Thu, Jan 8, 2009 at 2:04 PM, Joel Truher <joel@truher.org> wrote:
> > I'm curious about this topic, Rahul, since I encountered it as well.
> >
> > The solution you suggest seems to me to be external to SCXML, and it
> > made me wonder if this sort of templating ought to be supported more
> > natively, or if there's a design argument to exclude it?
> >
> > I searched for a way to avoid templating, since it does seem foreign
> > to the state-chart idea (as well as other implementations of it, e.g.
> > Boost.Statechart).  I eventually succeeded, and I think my design is
> > better for it.  So maybe the exclusion of templates is a good design
> > choice.  Is that how you see it?
> >
> <snip/>
>
> It has been a conscious design choice so far to say that ensuring the
> uniqueness of IDs is the responsibility of the document author in all
> cases. Some solutions were considered IIRC to automatically introduce
> some sort of scoping or ID-munging to get around this issue of src'ing
> in the same document multiple times, but none were found to be
> satisfactory. Given that SCXML will very likely be used to design
> systems that are distributed, reactive and heterogeneous it was
> considered that its not a stretch to say use a dynamic URL here. The
> advantage of the example approach outlined here is that it gives the
> author complete control over how the IDs will be treated every time,
> and these are therefore predictable at design-time.
>
> I'm not sure if you are saying you avoid templating or use some other
> authoring pattern (if you'd like to share with all users, you can
> create a wiki page if you want). The wiki is here [1] and open to all
> (after a quick registration).
>
> -Rahul
>
> [1] http://wiki.apache.org/commons/SCXML/HomePage
>
>
>
> >
> > Joel
> >
> >
> > On Thu, Jan 8, 2009 at 10:41 AM, Rahul Akolkar <rahul.akolkar@gmail.com>
> wrote:
> >> On Thu, Jan 8, 2009 at 8:37 AM, Lee, Cheryl - ES/RDR -Gil
> >> <Cheryl.Lee@itt.com> wrote:
> >>> Thanks for your response.
> >>>
> >>> I'm not familiar with using a JSP (assume are we talking Java Server
> >>> Pages?) in the context of my application.  Mine is a stand-alone Java
> >>> GUI application using Swing with the state machine to respond to user
> >>> actions and other system actions.  Is there a way to use a dynamic
> >>> resource without using JSP?  If so, what would an example look like?
> >>>
> >> <snip/>
> >>
> >> It doesn't have to be a JSP, it was used for illustration. All we're
> >> saying is use a dynamic URL to tweak the IDs as needed. If everything
> >> has to be local (no servlet container, web server or any type of
> >> remote service access), then it may not be straightforward to create
> >> such a dynamic URL (custom protocol handlers for the java.net.URL
> >> architecture or other fancy footwork perhaps).
> >>
> >> -Rahul
> >>
> >>
> >>> Thanks a bunch,
> >>> Cheryl
> >>>
> <snap/>
>
> ---------------------------------------------------------------------
> To unsubscribe, e-mail: user-unsubscribe@commons.apache.org
> For additional commands, e-mail: user-help@commons.apache.org
>
>

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