Hi, I am using the "forall" keyword on the LHS and it seems the condition is satisfied when there is nothing to match. e.g. "forall (MyClass(value == "test"))" fires the rule when there is no MyClass() object in the working memory. My workaround is add "exists (MyClass())". Also, same for "not (exists (MyClass(value != "test")))", but I can understand this, as there is no MyClass() object, it doesn't exists and hence "not" gives true. But for "forall", it doesnt sound right to me. I wonder if "forall" is actually implemented as "not exists" in the engine. Please clarify if this is by design or a bug. Build is 5.5.0.Final Thank you -- View this message in context: http://drools.46999.n3.nabble.com/forall-is-satisfied-when-there-is-nothi... Sent from the Drools: User forum mailing list archive at Nabble.com. _______________________________________________ rules-users mailing list rules-users(a)lists.jboss.org https://lists.jboss.org/mailman/listinfo/rules-users

First order logic does permit empty domains, and the universal quantifier evaluates to true in that case. Davide On 01/07/2014 11:11 AM, Wolfgang Laun wrote: > It is true that first-order logic usually assumes that the domain of a > formula > be a nonempty set. As so often, things aren't quite so simple when > formulae > are evaluated on a computer. What should be done in this case, forall > with an empty domain? Throw an exception? Not very convenient, since > there's no reasonable way of handling exceptions thrown on the LHS. > Return false? That doesn't make sense, because you can't inspect what > isn't there. Return true? If it isn't false - what else? > > If first-order logic does permit empty domains, it must be treated as > a special case. > > -W > > On 07/01/2014, Sonata <plz.write.to(a)gmail.com&gt; wrote: >> Hi, I am using the "forall" keyword on the LHS and it seems the >> condition >> is >> satisfied when there is nothing to match. e.g. "forall (MyClass(value == >> "test"))" fires the rule when there is no MyClass() object in the >> working >> memory. My workaround is add "exists (MyClass())". >> >> Also, same for "not (exists (MyClass(value != "test")))", but I can >> understand this, as there is no MyClass() object, it doesn't exists and >> hence "not" gives true. >> >> But for "forall", it doesnt sound right to me. I wonder if "forall" is >> actually implemented as "not exists" in the engine. >> >> Please clarify if this is by design or a bug. Build is 5.5.0.Final >> >> Thank you >> >> >> >> -- >> View this message in context: >> http://drools.46999.n3.nabble.com/forall-is-satisfied-when-there-is-nothi... >> Sent from the Drools: User forum mailing list archive at Nabble.com. >> _______________________________________________ >> rules-users mailing list >> rules-users(a)lists.jboss.org >> https://lists.jboss.org/mailman/listinfo/rules-users >> > _______________________________________________ > rules-users mailing list > rules-users(a)lists.jboss.org > https://lists.jboss.org/mailman/listinfo/rules-users > _______________________________________________ rules-users mailing list rules-users(a)lists.jboss.org https://lists.jboss.org/mailman/listinfo/rules-users

Indeed it is true by convention, see also http://en.wikipedia.org/wiki/Universal_quantification The only other alternative is to deprecate quantifiers altogether ;)

Davide On 01/07/2014 11:53 AM, Wolfgang Laun wrote: > Please comment on > http://en.wikipedia.org/wiki/First-order_logic#Empty_domains > -W > > On 07/01/2014, Davide Sottara <dsotty(a)gmail.com&gt; wrote: >> First order logic does permit empty domains, and the universal >> quantifier evaluates to true in that case. >> Davide >> >> On 01/07/2014 11:11 AM, Wolfgang Laun wrote: >>> It is true that first-order logic usually assumes that the domain of a >>> formula >>> be a nonempty set. As so often, things aren't quite so simple when >>> formulae >>> are evaluated on a computer. What should be done in this case, forall >>> with an empty domain? Throw an exception? Not very convenient, since >>> there's no reasonable way of handling exceptions thrown on the LHS. >>> Return false? That doesn't make sense, because you can't inspect what >>> isn't there. Return true? If it isn't false - what else? >>> >>> If first-order logic does permit empty domains, it must be treated as >>> a special case. >>> >>> -W >>> >>> On 07/01/2014, Sonata <plz.write.to(a)gmail.com&gt; wrote: >>>> Hi, I am using the "forall" keyword on the LHS and it seems the >>>> condition >>>> is >>>> satisfied when there is nothing to match. e.g. "forall (MyClass(value >>>> == >>>> "test"))" fires the rule when there is no MyClass() object in the >>>> working >>>> memory. My workaround is add "exists (MyClass())". >>>> >>>> Also, same for "not (exists (MyClass(value != "test")))", but I can >>>> understand this, as there is no MyClass() object, it doesn't exists >>>> and >>>> hence "not" gives true. >>>> >>>> But for "forall", it doesnt sound right to me. I wonder if "forall" is >>>> actually implemented as "not exists" in the engine. >>>> >>>> Please clarify if this is by design or a bug. Build is 5.5.0.Final >>>> >>>> Thank you >>>> >>>> >>>> >>>> -- >>>> View this message in context: >>>> http://drools.46999.n3.nabble.com/forall-is-satisfied-when-there-is-nothi... >>>> Sent from the Drools: User forum mailing list archive at Nabble.com. >>>> _______________________________________________ >>>> rules-users mailing list >>>> rules-users(a)lists.jboss.org >>>> https://lists.jboss.org/mailman/listinfo/rules-users >>>> >>> _______________________________________________ >>> rules-users mailing list >>> rules-users(a)lists.jboss.org >>> https://lists.jboss.org/mailman/listinfo/rules-users >>> >> _______________________________________________ >> rules-users mailing list >> rules-users(a)lists.jboss.org >> https://lists.jboss.org/mailman/listinfo/rules-users >> > _______________________________________________ > rules-users mailing list > rules-users(a)lists.jboss.org > https://lists.jboss.org/mailman/listinfo/rules-users > _______________________________________________ rules-users mailing list rules-users(a)lists.jboss.org https://lists.jboss.org/mailman/listinfo/rules-users

laune wrote > On 08/01/2014, Davide Sottara &lt; > dsotty@ > &gt; wrote: >> Indeed it is true by convention, see also >> http://en.wikipedia.org/wiki/Universal_quantification >> >> The only other alternative is to deprecate quantifiers altogether ;) >> > Ah, *by convention*, yes. One should exercise some care with "vacuous > truths", however. Uttering statements such as "All the little green > men in my room are from Mars" may not brand you as a liar, but you > could be called "batty". ;-) > > -W > >> Davide >> Guys, I am not sure whats the arguing point here. But from a programmer's and a Drools engine user's perspective, I would find forall(void) to return false more convenient. Say for example: forall(Cloth(dried)) then collect() Programmingwise, I dont want to execute collect() when the Cloth() collection is empty and do extra null checkings. Performancewise, I also dont want to call collect() when there is actually nothing to perform. In terms of semantics, I purposely want to check for isDried is true, it is easy to realize from the LHS that I only want to collect clothes when all of them are dried, rather than I want to collect also empty air. On the contrary, if I really want to do the RHS even though the LHS has nothing, I would have written: not( exists( Customer() ) ) then takeanap() and I will not write forall( Customer(status == "gone") ) then takeanap() because the latter needs to do extra property visits and string comparisons and it is so indirect to express what I really want. I really dont care what the mathematical definition is unless you tell me Drools is a math. engine rather than a rule engine to express business requirements. That, I should update myself and take a step back to look at Drools again. Just my own opinion :P -- View this message in context: http://drools.46999.n3.nabble.com/forall-is-satisfied-when-there-is-nothi... Sent from the Drools: User forum mailing list archive at Nabble.com. _______________________________________________ rules-users mailing list rules-users(a)lists.jboss.org https://lists.jboss.org/mailman/listinfo/rules-users

Davide Sottara wrote > This is actually the way it is implemented internally, and no, I don't > think that > it would be appropriate to change it. Yes I agree, so may be we users actually do not need forall, but *every*, which just means forall and exists Now, look at it again: when every Cloth( dried ) then collect() See how pretty it is, simple beauty, fully expressive, just like a sentence :) as oppose to when forall Cloth ( dried ) AND exists Cloth() then collect() X( Nah, I guess people can live with that :P -- View this message in context: http://drools.46999.n3.nabble.com/forall-is-satisfied-when-there-is-nothi... Sent from the Drools: User forum mailing list archive at Nabble.com. _______________________________________________ rules-users mailing list rules-users(a)lists.jboss.org https://lists.jboss.org/mailman/listinfo/rules-users

laune wrote > On 08/01/2014, Davide Sottara &lt; > dsotty@ > &gt; wrote: >> Indeed it is true by convention, see also >> http://en.wikipedia.org/wiki/Universal_quantification >> >> The only other alternative is to deprecate quantifiers altogether ;) >> > > Ah, *by convention*, yes. One should exercise some care with "vacuous > truths", however. Uttering statements such as "All the little green > men in my room are from Mars" may not brand you as a liar, but you > could be called "batty". ;-) > > -W > >> Davide >> Guys, I am not sure whats the arguing point here. But from a programmer's and a Drools engine user's perspective, I would find forall(void) to return false more convenient. Say for example: forall(Cloth(dried)) then collect() Programmingwise, I dont want to execute collect() when the Cloth() collection is empty and do extra null checkings. Performancewise, I also dont want to call collect() when there is actually nothing to perform. In terms of semantics, I purposely want to check for isDried is true, it is easy to realize from the LHS that I only want to collect clothes when all of them are dried, rather than I want to collect also empty air. On the contrary, if I really want to do the RHS even though the LHS has nothing, I would have written: not( exists( Customer() ) ) then takeanap() and I will not write forall( Customer(status == "gone") ) then takeanap() because the latter needs to do extra property visits and string comparisons and it is so indirect to express what I really want. I really dont care what the mathematical definition is unless you tell me Drools is a math. engine rather than a rule engine to express business requirements. That, I should update myself and take a step back to look at Drools again. Just my own opinion :P -- View this message in context: http://drools.46999.n3.nabble.com/forall-is-satisfied-when-there-is-nothi... Sent from the Drools: User forum mailing list archive at Nabble.com. _______________________________________________ rules-users mailing list rules-users(a)lists.jboss.org https://lists.jboss.org/mailman/listinfo/rules-users

On 08/01/2014, Davide Sottara <dsotty(a)gmail.com&gt; wrote: > Indeed it is true by convention, see also > http://en.wikipedia.org/wiki/Universal_quantification > > The only other alternative is to deprecate quantifiers altogether ;) > Ah, *by convention*, yes. One should exercise some care with "vacuous truths", however. Uttering statements such as "All the little green men in my room are from Mars" may not brand you as a liar, but you could be called "batty". ;-) -W > Davide > > On 01/07/2014 11:53 AM, Wolfgang Laun wrote: >> Please comment on >> http://en.wikipedia.org/wiki/First-order_logic#Empty_domains >> -W >> >> On 07/01/2014, Davide Sottara <dsotty(a)gmail.com&gt; wrote: >>> First order logic does permit empty domains, and the universal >>> quantifier evaluates to true in that case. >>> Davide >>> >>> On 01/07/2014 11:11 AM, Wolfgang Laun wrote: >>>> It is true that first-order logic usually assumes that the domain of a >>>> formula >>>> be a nonempty set. As so often, things aren't quite so simple when >>>> formulae >>>> are evaluated on a computer. What should be done in this case, forall >>>> with an empty domain? Throw an exception? Not very convenient, since >>>> there's no reasonable way of handling exceptions thrown on the LHS. >>>> Return false? That doesn't make sense, because you can't inspect what >>>> isn't there. Return true? If it isn't false - what else? >>>> >>>> If first-order logic does permit empty domains, it must be treated as >>>> a special case. >>>> >>>> -W >>>> >>>> On 07/01/2014, Sonata <plz.write.to(a)gmail.com&gt; wrote: >>>>> Hi, I am using the "forall" keyword on the LHS and it seems the >>>>> condition >>>>> is >>>>> satisfied when there is nothing to match. e.g. "forall (MyClass(value >>>>> == >>>>> "test"))" fires the rule when there is no MyClass() object in the >>>>> working >>>>> memory. My workaround is add "exists (MyClass())". >>>>> >>>>> Also, same for "not (exists (MyClass(value != "test")))", but I can >>>>> understand this, as there is no MyClass() object, it doesn't exists >>>>> and >>>>> hence "not" gives true. >>>>> >>>>> But for "forall", it doesnt sound right to me. I wonder if "forall" is >>>>> actually implemented as "not exists" in the engine. >>>>> >>>>> Please clarify if this is by design or a bug. Build is 5.5.0.Final >>>>> >>>>> Thank you >>>>> >>>>> >>>>> >>>>> -- >>>>> View this message in context: >>>>> http://drools.46999.n3.nabble.com/forall-is-satisfied-when-there-is-nothi... >>>>> Sent from the Drools: User forum mailing list archive at Nabble.com. >>>>> _______________________________________________ >>>>> rules-users mailing list >>>>> rules-users(a)lists.jboss.org >>>>> https://lists.jboss.org/mailman/listinfo/rules-users >>>>> >>>> _______________________________________________ >>>> rules-users mailing list >>>> rules-users(a)lists.jboss.org >>>> https://lists.jboss.org/mailman/listinfo/rules-users >>>> >>> _______________________________________________ >>> rules-users mailing list >>> rules-users(a)lists.jboss.org >>> https://lists.jboss.org/mailman/listinfo/rules-users >>> >> _______________________________________________ >> rules-users mailing list >> rules-users(a)lists.jboss.org >> https://lists.jboss.org/mailman/listinfo/rules-users >> > _______________________________________________ > rules-users mailing list > rules-users(a)lists.jboss.org > https://lists.jboss.org/mailman/listinfo/rules-users > _______________________________________________ rules-users mailing list rules-users(a)lists.jboss.org https://lists.jboss.org/mailman/listinfo/rules-users