In truth, I'd don't really understand why the EL spec is not allowing the null value through. Unless it detects that a primitive value is required by the assignment, it has no right trying to interpret a null value as anything other than null. In that sense, it has nothing to do with expected types, especially when the value is null. -Dan On Tue, Oct 13, 2009 at 5:03 PM, Martin Marinschek <mmarinschek(a)apache.org&gt;wrote: > Hi all, > > I don't know if we discussed this already, but today the coercion > issue in the EL made me loose a few hours again. I've had this before, > but seemingly forgot about it - time to follow up on this. > > The issue in short: collapsed="#{bb.collapsed}" > > Boolean getCollapsed() { > return null; > } > > will lead to a value of "false" for the collapsed attribute if > java.lang.Boolean is set as the expected type of the corresponding > value-expression, according to the EL spec. Hrmmpf. > > You can read more in this blog-entry: > http://www.irian.at/blog/blogid/unifiedElCoercion/#unifiedElCoercion > > Are we going to say that for JSF 2.0 Facelets we will never set the > expected-type? Or are existing EL implementations not following the > spec and this is not a problem in reality (however, at least the one > that I use does follow the spec)? > > regards, > > Martin > > -- > > http://www.irian.at > > Your JSF powerhouse - > JSF Consulting, Development and > Courses in English and German > > Professional Support for Apache MyFaces > -- Dan Allen Senior Software Engineer, Red Hat | Author of Seam in Action Registered Linux User #231597 http://mojavelinux.com http://mojavelinux.com/seaminaction http://www.google.com/profiles/dan.j.allen

Right, I agree with you. I'm saying that this is silly behavior. Either it should leave null alone or choose to convert it only *if* the expected-type is a primitive (in this case a primitive boolean).

> Right, I agree with you. I'm saying that this is silly behavior. > Either it should leave null alone or choose to convert it only *if* > the expected-type is a primitive (in this case a primitive boolean). +1

On Fri, Oct 16, 2009 at 4:49 AM, Martin Marinschek <mmarinschek(a)apache.org&gt; wrote: > > >> Right, I agree with you. I'm saying that this is silly behavior. > >> Either it should leave null alone or choose to convert it only *if* > >> the expected-type is a primitive (in this case a primitive boolean). > > > > +1 > > I can certainly do so - however, I don't think that I will get any > response amounting to this being fixed. > > Do you guys think we should fix this in the JSF implementations, by > passing in java.lang.Object to the underlying EL expression, and doing > the coercion ourselves? We - most of the time - wrap the > value-expression in a Facelets TagValueExpression anyways, so we could > certainly do this! If the RI team agrees, I can certainly give word > out to the MyFaces team discussing if that is reasonable to us as > well. I think it would be more reasonable to try to handle this ourselves. As you mentioned, there is already a layer in between, so it's not like this is a one off solution. To be honest, I'm on the fence, but I also want to get rid of annoyances sooner rather than later and that makes me lean towards a fix in JSF. -Dan -- Dan Allen Senior Software Engineer, Red Hat | Author of Seam in Action Registered Linux User #231597 http://mojavelinux.com http://mojavelinux.com/seaminaction http://www.google.com/profiles/dan.j.allen

Hi Lincoln, On 10/16/09, Lincoln Baxter, III <lincolnbaxter(a)gmail.com&gt; wrote: > I'm concerned... what implications does this have? Why is this > something we should be fixing? This is a potentially dangerous thing > to be tampering with IMO. I think I explained why this is a problem - it is against the "principle of least suprise": I don't expect, if I pass a null into an EL expression in the managed-bean, to have a "false" coming out in the attribute in my component. I expect the component author to be able to pick a sensible default value if I pass null. Same with a string - I don't expect an empty string if I pass null. What happens now is that people are not passing nulls from EL expressions, instead picking a default value themselves. It is most certainly not a decent solution to place the burden of picking default values on the web-app developer... regards, Martin

The differences between JSF EL and JSTL EL behavior has already caught me off guard, so there's another side to the principle of least surprise that needs to be considered.