Re: [mdt-papyrus.dev] [Papyrus] Risk of regressions for Papyrus diagrams

[Vincent Hémery <vincent.hemery@xxxxxxxxxxxxxx>]

Dear Ansgar,

I do not think this is related at all.
In fact, there is no bug in the generated code.
Only the customized code using the generated code is impacted, because 
one can call the same methods Polyline and PolylineShape whereas the 
result is not the same.

The problem appears only when you update the points manually. Because 
Polyline and PolylineShape do not impose the same contract.

In my case, I knew the definitive shape of my Polyline only at runtime. 
So I was initializing the Polyline with fewer points than needed, 
knowing that I would update it later. When manually adding the necessary 
points, in Polyline, the size was adjusting automatically.

On the other hand, PolylineShape relies on the #setBounds method and 
will not display points which are out of its bounds. Hence, adding 
points out of its current bounds will lead to a truncated or to an 
invisible line.

Add to this the fact that the bounds given by the #setBounds method were 
taken in account for the layout of figures, and you will know all the 
impacts switching from Polyline to PolylineShape can have on a 
customized code.
This is the reason why I was talking about eventual layout problems, 
since Polyline could trick the layout, drawing a big image and taking 
small place regarding the layout (whereas PolylineShape can't cheat).

Best regards,

Vincent.

Le 05/08/2010 17:58, Ansgar Radermacher a écrit :
> Dear Vincent,
>
> I might be totally wrong, but since it helps to set the bounds
> explicitly, it may be related to the fixed GMF bug
> https://bugs.eclipse.org/bugs/show_bug.cgi?id=319191
> (https://bugs.eclipse.org/bugs/show_bug.cgi?id=319348)
> As far as I know, GMF 2.3.1 is not yet used by Papyrus.
>
> Best regards
>
> Ansgar
>
> PS: Regression is a big problem in general, as we do not have automated
> tests ...
>
>
> On 08/05/2010 05:45 PM, Vincent Hémery wrote:
> > Hello everyone,
> >
> > There has been changes in gmfgen files generated by GMF, from the M7.
> >
> > One of them may eventually impact any diagram using generated figures
> > from a .gmfgraph file with Polyline objects.
> >
> > It has already caused regressions in Activity diagram (corrected
> > before the freeze) and in Sequence diagram (still in correction).
> >
> >
> >
> > Symptoms :
> > A polyline is no longer displayed. Especially in case it is supposed
> > to have changed.
> >
> > Reasons :
> > In the final generated code, Polyline objects used to generate
> > org.eclipse.draw2d.Polyline instances. Whereas since the M7,
> > org.eclipse.draw2d.PolylineShape instances are generated instead.
> >
> > Corrections :
> > In case you were updating the polyline's points manually, it should no
> > longer appear. To make it reappear, you must in addition call the
> > method org.eclipse.draw2d.Figure.setBounds(Rectangle) with the
> > appropriate bounds.
> > In case your Polyline used to take the whole place, there should be no
> > more problem if you set the correct bounds (do not forget to take
> > figure's insets in account).
> > On the other hand, if your parent figure was containing other figures,
> > you may face layout problems. In such a case, do not hesitate to
> > consider changing the parent layout.
> >
> >
> > I Hope your diagrams won't be impacted.
> >
> > Best regards,
> >
> > Vincent Hémery.
> > _______________________________________________
> > mdt-papyrus.dev mailing list
> > mdt-papyrus.dev@xxxxxxxxxxx
> > https://dev.eclipse.org/mailman/listinfo/mdt-papyrus.dev