David, Thanks for the changes in ManyToMany. I am testing now. The type parameters did seem to me redundant, and compiler hostile.
Naftoli, You are correct that the finished product will have preset roles that can be chosen from a select box. But there is a use-case for allowing on-the-fly creation of children in a many-to-many scenario. On Aug 8, 10:35 pm, Naftoli Gugenheim <[email protected]> wrote: > Thanks, David. > Glenn, as far as the issue of adding unsaved children, I was thinking about > it. If the same Role can be referenced by multiple users, why are you > creating Roles at the same time that you are adding them to a User? Shouldn't > there be one screen to manage Roles and the User screen would only allow you > to add existing Roles? > > ------------------------------------- > > David Pollak<[email protected]> wrote: > > There are dependent types that mirror the parameterized types for Mapper, > KeyedMapper, etc. > I've updated ManyToMany to use the dependent types... it should eliminate > the need to have the type parameters. If it breaks thing, please revert the > changes. > > On Thu, Aug 6, 2009 at 9:55 AM, Naftoli Gugenheim <[email protected]>wrote: > > > In your use case it can only be Long and User. But there has to be a type > > parameter because other people might have a String key and a Request mapper. > > And the contents of ManyToMany have to be type safe to enforce their > > consistency with however the user of ManyToMany wants them to use it. So I > > think what you want is that the compiler should figure out that since you're > > using it with a KeyedMapper[Long, User], which is ManyToMany's T <: > > KeyedMapper[K,T], and ManyToMany extends KeyedMapper[K,T] and it has a self > > type of T, and therefore it would be a compiler error to write anything > > besides Long and User, therefore you must want to use Long and User and they > > should be inferred. > > I could certainly hear that, although I'm not much of a type theorist. So > > feel free to open an enhancement ticket on Scala's Trac! :) > > If I understood correctly... > > > On Thu, Aug 6, 2009 at 12:43 PM, glenn <[email protected]> wrote: > > >> Naftoli, > > >> As I said at the outset, this is really beyond my expertise. But I > >> think it's too > >> broad and maybe, even unnecessary. In my example, K can only be one > >> type, > >> Long, and T can only be of type User. Anything else, and the compiler > >> can't be > >> guaranteed to catch it, but try running the application and you get > >> things like > >> stack overflow errors. > > >> How you would fix this is beyond me. All you can really do, at this > >> point, is > >> make sure to include these kinds of restrictions in the docs. > > >> Glenn... > > >> On Aug 6, 9:33 am, Naftoli Gugenheim <[email protected]> wrote: > >> > It's too broad, or it's too restrictive/unnecessary? I'm confused, you > >> seem > >> > to imply both.If it's too broad, tell me why. But if you want to know > >> why it > >> > needs the type parameter, it's because it has to use the type parameter > >> in > >> > its implentation -- just count how many times its source code uses it! > >> And > >> > if you want to know why the compiler can't infer it or be told to infer > >> > it... that's another issue. > >> > But if passing a type that creates a conflicting inheritance causes a > >> > compiler crash, that's not something I can help. :) > >> > Regards. > > >> > On Thu, Aug 6, 2009 at 12:26 PM, glenn <[email protected]> wrote: > > >> > > Naftoli, > > >> > > At the risk of discussing something obviously beyond my pay grade, > >> > > isn't the real issue Scala traits > >> > > and the use of parameterized types. The ManyToMany trait is defined > >> > > as: > > >> > > trait ManyToMany[K,T<:KeyedMapper[K, T]].... > > >> > > But this isn't really correct,is it, the parameter is too broad, and > >> > > that leads to a lot of the confusion and results in the need for > >> > > unnecessary documentation. When you think about it, why all the > >> > > duplication? Why do I need to > >> > > write my User entity as: > > >> > > class User extends MegaProtoUser[User] with ManyToMany[Long,User] > > >> > > when we all know that this would be much cleaner: > > >> > > class User extends MegaProtoUser[User] with ManyToMany. > > >> > > But traits aren't like interfaces. They have implementation and would > >> > > need to know something about > >> > > the parent class they are attached to - and how would you accomplish > >> > > that (reflection, maybe). This is > >> > > more a Scala issue than a Lift one, I think. > > >> > > Glenn... > > >> > > On Aug 5, 8:24 pm, Naftoli Gugenheim <[email protected]> wrote: > >> > > > Building causes a stack overflow? > >> > > > So the question is, is it the resident compiler or plain scalac also > >> > > crashes? Or just the presentation compiler? What do you see in the > >> error log > >> > > view or file? > >> > > > I get compiler crashes very often when doing fancy mapper type > >> related > >> > > tricks. > > >> > > > ------------------------------------- > > >> > > > glenn<[email protected]> wrote: > > >> > > > Naftoli, > > >> > > > Hate to do this to you, but I'm getting the following error using > >> > > > ManyToMany for Users to Roles: > > >> > > > Message: java.lang.RuntimeException: Broken join > >> > > > scala.Predef$.error(Predef.scala:76) > > >> > > net.liftweb.mapper.ManyToMany$MappedManyToMany$$anonfun$children$1$ > >> > > > $anonfun$apply$1.apply(ManyToMany.scala:54) > > >> > > net.liftweb.mapper.ManyToMany$MappedManyToMany$$anonfun$children$1$ > >> > > > $anonfun$apply$1.apply(ManyToMany.scala:54) > >> > > > net.liftweb.util.EmptyBox.openOr(Box.scala:372) > > >> net.liftweb.mapper.ManyToMany$MappedManyToMany$$anonfun$children > >> > > > $1.apply(ManyToMany.scala:54) > > >> net.liftweb.mapper.ManyToMany$MappedManyToMany$$anonfun$children > >> > > > $1.apply(ManyToMany.scala:54) > >> > > > scala.List.map(List.scala:812) > >> > > > net.liftweb.mapper.ManyToMany$MappedManyToMany.children > >> > > > (ManyToMany.scala:54) > >> > > > net.liftweb.mapper.ManyToMany$MappedManyToMany.elements > >> > > > (ManyToMany.scala:96) > >> > > > scala.Seq$class.flatMap(Seq.scala:293) > >> > > > net.liftweb.mapper.ManyToMany$MappedManyToMany.flatMap > >> > > > (ManyToMany.scala:44) > > >> > > > def edit(ns: NodeSeq): NodeSeq = { > >> > > > val theUser = view.entity > >> > > > val addRole = TheBindParam("insert", view.snippet.link("edit", > >> () > >> > > > => theUser.roles += new Role, Text(S?("Add Role")))) > > >> > > > bind("user", ns, > >> > > > "firstname" -> text(theUser.firstName.is, > >> theUser.firstName > >> > > > (_), ("size","20")), > >> > > > "lastname" -> text(theUser.lastName.is > >> ,theUser.lastName(_), > >> > > > ("size", "30")), > >> > > > "roles" -> theUser.roles.flatMap{role => > >> > > > bind("role", ns, > >> > > > "name" -> role.name.toForm, > >> > > > "remove" -> SHtml.submit(S?("Remove"), ()=> > >> > > > theUser.roles -= role) > >> > > > ) > >> > > > }, > >> > > > addRole, > >> > > > "submit" -> SHtml.submit(S?("Save"), ()=>view.save) > >> > > > ) > >> > > > } > > >> > > > The offending code seems to be the line: "roles" -> > >> > > > theUser.roles.flatMap{.... > >> > > > in the above bind method when I click on the addRole link. > > >> > > > Here's my User class: > > >> > > > class User extends MegaProtoUser[User] with ManyToMany[Long,User]{ > >> > > > def getSingleton = User // what's the "meta" server > > >> > > > object roles > >> > > > extends MappedManyToMany(UserRole, UserRole.user, UserRole.role, > >> > > > Role) > > >> > > > } > > >> > > > What am I doing wrong? You can see how difficult it is to slog > >> through > >> > > > this code, let alone just > >> > > > trying to explain the problem so I can get help. > > >> > > > On Aug 5, 9:57 am, Naftoli Gugenheim <[email protected]> wrote: > > >> > > > > I'll try. > >> > > > > By the way, as per my correction, you can implement list the > >> regular > >> > > way without ModelView, and just use ModelSnippet's load function in > >> your > >> > > edit link or button, passing it the User instance. > > >> > > > > ------------------------------------- > > >> > > > > glenn<[email protected]> wrote: > > >> > > > > Naftoli, > > >> > > > > I fixed my code per your comments and now I can edit and remove > >> users > >> > > > > from a list, as long as I populate the list with > >> > > > > ModelView instances, as you said. As for the docs, this step was > >> not > >> > > > > clear to me at all. I just assumed that the list was > >> > > > > just populated with User entities and the view in the ModelSnippet > >> was > >> > > > > instantiated with the selected User on each request. > > >> > > > > It sounds like your plate is pretty full, so I won't expect much, > >> but > >> > > > > sometime soon, could you provide an example, or improved > >> > > > > docs, for using TableEditor and its related ItemsList trait. > > >> > > > > Thanks for all. > > >> > > > > Glenn... > > >> > > > > On Aug 5, 9:18 am, Naftoli Gugenheim <[email protected]> > >> wrote: > > >> > > > > > Correction: ModelSnippet.load takes the actual Mapper instance, > >> not > >> > > the ModelView wrapper. > > >> > > > > > ------------------------------------- > > >> > > > > > Naftoli Gugenheim<[email protected]> wrote: > > >> > > > > > To answer your immediate question, the listing should not refer > >> to > >> > > the snippet's view but new ModelView instances for each entity. Then > >> > > editAction is shorthand for the snippet's link method with a callback > >> to > >> > > call load on the ModelView. To set the snippet's view's entity, either > >> call > >> > > load on the snippet with the other ModelView, or call load on the > >> other > >> > > ModelView (or just set its entity directly). > >> > > > > > As far as documentation, please tell me what scaladocs need what > >> > > clarification. Thanks. > > >> > > > > > ------------------------------------- > > >> > > > > > glenn<[email protected]> wrote: > > >> > > > > > Naftoli, > > >> > > > > > Functional programming systems are notoriously difficult to > >> document. > > ... > > read more » --~--~---------~--~----~------------~-------~--~----~ You received this message because you are subscribed to the Google Groups "Lift" group. To post to this group, send email to [email protected] To unsubscribe from this group, send email to [email protected] For more options, visit this group at http://groups.google.com/group/liftweb?hl=en -~----------~----~----~----~------~----~------~--~---
