Eclipse Community Forums: OCL » [OCLinEcore] Invariant evaluation only after validation of another invariant

Help

Home

Home » Modeling » OCL » [OCLinEcore] Invariant evaluation only after validation of another invariant

Show: Today's Messages :: Show Polls :: Message Navigator

[OCLinEcore] Invariant evaluation only after validation of another invariant [message #1014783]

Tue, 26 February 2013 13:11

Cedric Moonen

Messages: 274
Registered: August 2009

Senior Member

Hello,

I have several constraints on an object of my meta-model but for some of them, the invariant should be evaluated only if another invariant was evaluated positively (didn't produce an error).
In fact, I first evaluate certain condition and if they are not fulfilled, I don't want to evaluate another invariant (it doesn't make sense in the situation).

Is it possible to do something like that ?

Thanks,
Cédric

Report message to a moderator

Re: [OCLinEcore] Invariant evaluation only after validation of another invariant [message #1014796 is a reply to message #1014783]

Tue, 26 February 2013 13:32

Ed Willink

Messages: 7655
Registered: July 2009

Senior Member

Hi Cedric

No, but thanks for reminding me of this use case.

A previous discussion suggested a special dependency syntax. After a bit
more thought it seemed that making a constraint available as a
property/operation would allow a dependent constraint to be written as

inv SimpleConstraint: ...
inv ComplexConstraint: SimpleConstraint implies ...

you could do this manually now by

def checkSimpleConstraint() : Boolean = ...
inv SimpleConstraint: checkSimpleConstraint()
inv ComplexConstraint: checkSimpleConstraint() implies ...

which argues for no change needed. Just requires a smarter evaluator to
do a little Common Subexpression Elimination to detect the dependencies.
Enhancement request raised as
https://bugs.eclipse.org/bugs/show_bug.cgi?id=401786.

Regards

Ed Willink

On 26/02/2013 13:11, Cedric Moonen wrote:
> Hello,
>
> I have several constraints on an object of my meta-model but for some
> of them, the invariant should be evaluated only if another invariant
> was evaluated positively (didn't produce an error). In fact, I first
> evaluate certain condition and if they are not fulfilled, I don't want
> to evaluate another invariant (it doesn't make sense in the situation).
>
> Is it possible to do something like that ?
>
> Thanks,
> Cédric

Report message to a moderator