Yes it has been a while :). Comments below.
On 8/14/2007 6:47 AM PST, Simon Lessard wrote:
Hello Pavitra,
It was a long time since I last heard from you :). See my comments below.
Regards,
~ Simon
On 8/13/07, *Pavitra Subramaniam * <[EMAIL PROTECTED]
<mailto:[EMAIL PROTECTED]>> wrote:
Hi Simon,
I like your recommendations below and I have a few suggestions.
1. I have seen use cases where a model can be non-sequential, iow,
stops
in the train can appear in a random order and all are enabled the
first
time the processTrain comes up. In such cases getNextStep() and
getPreviousStep() may not be relevant. So perhaps the current
ProcessModel should be renamed to SequentialProcessModel (or
OrderedProcessModel something similar).
True, hence should not be in base ProcessModel as Adam said. We can
provide a decorator to wrap process model to add previous/next step
support. That wrapper could be named SequentialProcessModel or
OrderedProcessModel
2. Suggest renaming isPreviousStepAvailable() and
isNextStepAvailable()
to hasPreviousStep() and hasNextStep().
Sounds better, but alas not EL compatible, so I would prefer my
original suggestion.
Yes, I totally forgot that it may not be EL compatible :(.
May also be useful to have a
'isCurrentStep()' method.
That method would return something like
getRowKey().equals(getFocusRowKey())?
Yes
3. The fact that you have methods getNextStep() and getPreviousStep(),
implies we can figure out if a previous step is visited, disabled etc.
by doing the following
if (hasPreviousStep())
{
if (getPreviousStep().isVisited())
...
}
Yep, could be useful for a page template with back/forward buttons.
4. MaxPathKeyProcessModel could be renamed to MaxVisitedProcessModel
although like Adam recommends this could be part of the
ProcessModel (or
the OrderedProcessModel) class itself.
Adam wanted to keep it in ProcessMenuModel (where it's right now).
The only reason why I would like to place that code in another class
is to make it more obvious for users. I do like the
MaxVisitedProcessModel name.
5. Ideally we want some of these methods to be part of a
'ProcessStepModel' because whether a step is visited or disabled or
active etc. pertains to a step and not to the train. Although the
immediate property of a step can be at the ProcessModel level or
at the
step level.
That part is more controversial. Currently I'm not aware of any model
class that force the type of its content. I agree that semantically
this should be the responsibility of the the step though. I don't know
about immediate, I would say step level as well. I wouldn't mind
creating a ProcessStep class though, but I don't know if it should be
ProcessStepModel as Model classes normally represent a collection in
JSF and this could confuse new users.
I agree. Perhaps this class should be an implementation detail rather
than defined in the model interface.
Thanks
pavitra
Thanks
Pavitra
On 8/13/2007 12:56 PM PST, Simon Lessard wrote:
> Yeah the maxPath part could be just left in ProcessMenuModel.
>
> Previous/Next is not for train component, but rather for page
> templates using both a train and previous/next step buttons. The
only
> problem I have with placing those in a sub class is reuse with
> multiple implementations of ProcessModel. However, while thinking
> about it, it could be possible using decorator pattern in a generic
> way to add previous/next support to an existing model. Therefore, I
> don't think I'll add those.
>
>
> Regards,
>
> ~ Simon
>
> On 8/13/07, *Adam Winer* <[EMAIL PROTECTED]
<mailto:[EMAIL PROTECTED]> <mailto: [EMAIL PROTECTED]
<mailto:[EMAIL PROTECTED]>>>
> wrote:
>
> D'oh. :) Quite true. Could we also eliminate the "MaxPath"
model
> part,
> and just merge that into ProcessMenuModel?
>
> W/regard to the previous/next: in theory, the train could run
> entirely off of those, by iterating both forward and backward.
> But that's not necessary. Anyway, I'd be OK (I think) with
> seeing those in the base model, since they are quite generically
> useful, but I haven't thought it through in detail or tried
applying
> it to the existing component set.
>
> -- Adam
>
>
> On 8/13/07, Simon Lessard <[EMAIL PROTECTED]
<mailto:[EMAIL PROTECTED]>
> <mailto:[EMAIL PROTECTED]
<mailto:[EMAIL PROTECTED]>>> wrote:
> > The real issue imho is more with the methods linked with
> previous/next step
> > management. I guess those could be easily placed in the
> subclass though as
> > the train itself won't use them.
> >
> > public abstract boolean isNextStepAvailable(); /* Could be
> optional or
> > maybe in a subclass, this would check if there's a step
before
> the current
> > one */
> > public abstract boolean isPreviousStepAvailable(); /* As
above */
> > public abstract Object getNextStep(); /* As above */
> > public abstract Object getPreviousStep(); /* As above */
> > Regards,
> >
> > ~ Simon
> >
> >
> > On 8/13/07, Simon Lessard <[EMAIL PROTECTED]
<mailto:[EMAIL PROTECTED]>
> <mailto:[EMAIL PROTECTED]
<mailto:[EMAIL PROTECTED]>>> wrote:
> > > Hmmm, isn't that what I suggested? Current class is
> ProcessMenuModel, the
> > new one does not include the "Menu" part.
> > >
> > >
> > >
> > > On 8/13/07, Adam Winer < [EMAIL PROTECTED]
<mailto:[EMAIL PROTECTED]>
> <mailto:[EMAIL PROTECTED] <mailto:[EMAIL PROTECTED]>>> wrote:
> > > > So... what if we left the current class alone, in
terms of
> its API and
> > > > name, and just exposed a new base class?
> > > >
> > > > -- Adam
> > > >
> > > >
> > > > On 8/13/07, Adam Winer <[EMAIL PROTECTED]
<mailto:[EMAIL PROTECTED]>
> <mailto:[EMAIL PROTECTED] <mailto:[EMAIL PROTECTED]>> > wrote:
> > > > > I've got some concern for backwards compatibility -
> this'll break
> > > > > some code on our end. I'm hoping Jeanne will have some
> comments
> > > > > too.
> > > > >
> > > > > -- Adam
> > > > >
> > > > >
> > > > > On 8/13/07, Martin Marinschek <
> [EMAIL PROTECTED]
<mailto:[EMAIL PROTECTED]>
<mailto:[EMAIL PROTECTED]
<mailto:[EMAIL PROTECTED]>> >
> wrote:
> > > > > > No voice means you can go ahead ;)
> > > > > >
> > > > > > regards,
> > > > > >
> > > > > > Martin
> > > > > >
> > > > > > On 8/13/07, Danny Robinson <
[EMAIL PROTECTED] <mailto:[EMAIL PROTECTED]>
> <mailto:[EMAIL PROTECTED]
<mailto:[EMAIL PROTECTED]>>> wrote:
> > > > > > > Sorry Simon, I have little/no experience with
this part of
> > Trinidad so can't
> > > > > > > comment. I trust your judgement, so you have my
vote
> if you need
> > it ;-)
> > > > > > >
> > > > > > >
> > > > > > > On 8/13/07, Simon Lessard
<[EMAIL PROTECTED] <mailto:[EMAIL PROTECTED]>
> <mailto: [EMAIL PROTECTED]
<mailto:[EMAIL PROTECTED]>> > wrote:
> > > > > > > > So I assume it would be +0 for everyone?
> > > > > > > >
> > > > > > > >
> > > > > > > >
> > > > > > > > On 8/10/07, Simon Lessard <
> [EMAIL PROTECTED] <mailto:[EMAIL PROTECTED]>
<mailto: [EMAIL PROTECTED]
<mailto:[EMAIL PROTECTED]>>> wrote:
> > > > > > > > > Hello everybody,
> > > > > > > > >
> > > > > > > > > Currently Trinidad includes a ProcessMenuModel
> class that
> > contains
> > > > > > > undesirable methods. The complete method list (not
> including
> > inherited ones)
> > > > > > > is:
> > > > > > > > >
> > > > > > > > >
> > > > > > > > > public boolean isImmediate()
> > > > > > > > > public boolean isReadOnly()
> > > > > > > > > public boolean isVisited()
> > > > > > > > > public void clearMaxPath()
> > > > > > > > > public void setMaxPathKey(Object maxPathKey)
> > > > > > > > > public Object getMaxPathKey()The methods
involving
> maxPathKey
> > are the
> > > > > > > ones annoying me the most. However, as it's part of
> the public API
> > we have
> > > > > > > to keep backward compatibility as much as possible.
> Note also that
> > > > > > > isReadOnly should not named that way as readOnly
> support was
> > removed from
> > > > > > > process classes in favor of disabled since a
readOnly
> link did not
> > make much
> > > > > > > sense.
> > > > > > > > >
> > > > > > > > > Anyway, I would rather have the following class
> structure and
> > method
> > > > > > > signatures:
> > > > > > > > >
> > > > > > > > > public abstract class ProcessModel
> > > > > > > > >
> > > > > > > > >
> > > > > > > > > public abstract boolean isDisabled();
> > > > > > > > > public abstract boolean isImmediate();
> > > > > > > > > public abstract boolean isVisited();
> > > > > > > > > public abstract boolean
isNextStepAvailable(); /*
> Could be
> > optional or
> > > > > > > maybe in a subclass, this would check if there's a
> step before the
> > current
> > > > > > > one */
> > > > > > > > > public abstract boolean
isPreviousStepAvailable();
> /* As above
> > */
> > > > > > > > > public abstract Object getNextStep(); /* As
above */
> > > > > > > > > public abstract Object getPreviousStep(); /* As
> above */public
> > class
> > > > > > > MaxPathKeyProcessModel extends ProcessModel /* Or a
> better name */
> > > > > > > > >
> > > > > > > > >
> > > > > > > > > Implements all methods using the old
ProcessMenuModel
> > [EMAIL PROTECTED]
> > > > > > > > > public class ProcessMenuModel extends
> MaxPathKeyProcessModel
> > > > > > > > >
> > > > > > > > >
> > > > > > > > > Empty class except for isReadOnly() that
should return
> > > > > > > super.isDisabled ()
> > > > > > > > > The structure above would clean up the Model
class
> that really
> > shouldn't
> > > > > > > contain very implementation specific code like
the max
> path key
> > algorithm
> > > > > > > and would allow us to add new ProcessModel classes
> with more
> > > > > > > functionalities. For example I had one using a mode
> for step
> > access right
> > > > > > > like: MAX_PLUS_NEXT, MAX, ANY, NEXT_ONLY, etc.
> > > > > > > > >
> > > > > > > > > The previous/next step methods could be
useful for
> page
> > templates since
> > > > > > > it would be possible to include the train in the
> header as well as
> > a
> > > > > > > previous and next step buttons in the page
footer in a
> generic way
> > using the
> > > > > > > very same process model. Note that we might have to
> also include
> > methods
> > > > > > > like isPreviousVisited(), isPreviousDisabled() and
> such to fully
> > support
> > > > > > > that.
> > > > > > > > >
> > > > > > > > > Opinions, suggestions?
> > > > > > > > >
> > > > > > > > >
> > > > > > > > > Regards,
> > > > > > > > >
> > > > > > > > > ~ Simon
> > > > > > > > >
> > > > > > > > >
> > > > > > > > >
> > > > > > > > >
> > > > > > > >
> > > > > > > >
> > > > > > >
> > > > > > >
> > > > > > >
> > > > > > > --
> > > > > > > Chordiant Software Inc.
> > > > > > > www.chordiant.com <http://www.chordiant.com>
<http://www.chordiant.com>
> > > > > >
> > > > > >
> > > > > > --
> > > > > >
> > > > > > http://www.irian.at
> > > > > >
> > > > > > Your JSF powerhouse -
> > > > > > JSF Consulting, Development and
> > > > > > Courses in English and German
> > > > > >
> > > > > > Professional Support for Apache MyFaces
> > > > > >
> > > > >
> > > >
> > >
> > >
> >
> >
>
>
