20 Oct
2011
20 Oct
'11
12:05 a.m.
Ok. I imagine that that lazy metrics' getter calculates the result each time
it is performed, isn't it?
In our Moose extension we have problems when that happens.
Martin
On Wed, Oct 19, 2011 at 6:06 PM, Mariano Martinez Peck <
marianopeck(a)gmail.com> wrote:
On Wed, Oct 19, 2011 at 11:02 PM, Tudor Girba <tudor(a)tudorgirba.com>wrote;wrote:
Hi,
Yes, they are getters.
I now solved the problem by marking the metrics as <derived> which means
that they do not get saved.
Good news!
This means that the tests are green?
This was needed in any case. But, still the error that we got in Fuel might
still be around.
Well, this is not a bug/error but instead a serializer limitation: if a
certain object of the graph changes while we are serializing it, there may
be errors.
Cheers,
Doru
On 19 Oct 2011, at 21:23, Mariano Martinez Peck wrote:
--
Mariano
http://marianopeck.wordpress.com
On Wed, Oct 19, 2011 at 12:43 PM, Mariano Martinez Peck <
marianopeck(a)gmail.com> wrote:
On Wed, Oct 19, 2011 at 8:05 AM, Tudor Girba <tudor(a)tudorgirba.com>
wrote:
Hi,
Interesting observation. Perhaps, the problem could come from the
following
situation:
We now introduced meta-described properties for MooseEntities that
compute the
values lazily.
I forgto to say something. I don't understand why while traversing the
graph
we could trigger that lazy computation. Are they getters? can you
imagen a why that we execute them (even without wanting) while traversing
the graph?
Furthermore, the values are not stored in instance variables, but in a
dictionary
in MooseEntityState.
Indeed, that sounds related to the problem. I really never touch a line
in the
Fuel extension for Fuel nor to the Moose model, so I can be asking
non-sense. Anyway, we have this method:
shouldIgnore: attribute withAll: values
"Copied from FMRepositoryVisitor>>ignoreProperty:withAll:"
^ values isEmpty or: [
attribute isDerived or: [
attribute type == FM3MetaDescription boolean and: [
values size == 1 and: [
values first == false ]]]]
or
attributes
^attributes ifNil: [ attributes := (repository metamodel
descriptionOf:
theClass) allAttributes reject: [ :anAttribute | anAttribute
isDerived ] ].
and things like that... so may question is , maybe there are some new
attributes/properties which should be considered as derived for example?
cheers
I am not sure how exactly you are traversing the object graph and what
you are
serializing. I thought that you go exclusively through the
meta-description. In this case, you should not have this issue at all. But,
if you get to the object level, things can change.
Cheers,
Doru
On 19 Oct 2011, at 05:47, Martin Dias wrote:
> Yes, I couldn't invest many time to the fix it. I will continue during
the week.
>
> Martin
>
> On Tue, Oct 18, 2011 at 5:40 PM, Mariano Martinez Peck <
marianopeck(a)gmail.com> wrote:
>
>
> On Tue, Oct 18, 2011 at 10:34 PM, Tudor Girba <tudor(a)tudorgirba.com>
wrote:
> Hi,
>
> Any news related to this topic?
>
>
> Both Martin and I could reproduce the problem but we couldn't fix it.
We
are a little bit busy because we are finishing the paper... If course
Fuel can serialize Float, so the problem is in another place.
> So far we think that there is some instVar
than changes while we are
serializing the graph. Like if at the beginning it was
0.0 but then
something different. So it looks like a instVar which is a float and changes
while we are serializing it. That's all we can get for the moment.
>
> Cheers
>
> Cheers,
> Doru
>
> On 15 Oct 2011, at 16:42, Tudor Girba wrote:
>
> > Thanks!
> >
> > Keep us posted,
> > Doru
> >
> >
> > On 15 Oct 2011, at 15:10, Martin Dias wrote:
> >
> >> Hi
> >> I am downloading and will look at it.
> >>
> >> Martín
> >>
> >> On Sat, Oct 15, 2011 at 4:50 AM, Stéphane Ducasse <
stephane.ducasse(a)inria.fr> wrote:
> >> Thanks doru for the report!
> >>
> >>
> >> Stef
> >>
> >> On Oct 15, 2011, at 9:48 AM, Tudor Girba wrote:
> >>
> >>> Hi Mariano, hi Martin,
> >>>
> >>> As you know, we are using Fuel for serializing objects based on
Fame descriptions. We recently added new properties to Fame that contain
number values that are not integer (e.g., Float), and now we get the problem
that Fuel does not seem to be able to serialize them.
> >>>
> >>> To reproduce the problem, you can download the latest moose image
and run the FuelMooseTests:
> >>>
http://ci.moosetechnology.org/job/moose-latest-dev/lastSuccessfulBuild/arti…
> >>>
> > >>>
You will see that there are some 44 errors due to looking up
values in the dictionary. Could please you take a look and see if the
problem is in Fuel or maybe somewhere else?
> >>>
> > >>>
Cheers,
> > >>> Doru
> >>>
> >>>
> > >>> --
> > >>> www.tudorgirba.com
> >>>
> > >>>
"It's not how it is, it is how we see it."
> >>>
> >>>
> > >>>
_______________________________________________
> > >>> Moose-dev mailing list
> > >>> Moose-dev(a)iam.unibe.ch
> > >>> https://www.iam.unibe.ch/mailman/listinfo/moose-dev
> > >>
> > >>
> > >> _______________________________________________
> > >> Moose-dev mailing list
> > >> Moose-dev(a)iam.unibe.ch
> > >> https://www.iam.unibe.ch/mailman/listinfo/moose-dev
> > >
> > > --
> > > www.tudorgirba.com
> > >
> > > "In a world where everything is moving ever faster,
> > > one might have better chances to win by moving slower."
> > >
> > >
> > >
> >
> > --
> > www.tudorgirba.com
> >
> > "The coherence of a trip is given by the clearness of the goal."
> >
> >
> >
> >
> >
> >
> >
> > --
> > Mariano
> > http://marianopeck.wordpress.com
> >
> >
>
> --
> www.tudorgirba.com
>
> "Speaking louder won't make the point worthier."
>
>
>
>
> --
> Mariano
> http://marianopeck.wordpress.com
>
>
>
>
> --
> Mariano
> http://marianopeck.wordpress.com
>
--
www.tudorgirba.com
"Not knowing how to do something is not an argument for how it cannot be
done."