Don't forget anonymous validators will carry a reference to the page (or whatever the outside class is) so they will be huge when serialized.
The way I see it: adding an Imodel to the formcomponent is cheaper then a hashmap or a minimap or whatever and it is 100% general use case so it should def be part of the framework. Building that context might be way overkill. 2) can already be done by providing your own IValidatorResourceKeyFactory - I created that patch for this exact year months ago. -Igor > -----Original Message----- > From: [EMAIL PROTECTED] > [mailto:[EMAIL PROTECTED] On Behalf Of > Juergen Donnerstag > Sent: Wednesday, October 12, 2005 12:13 PM > To: wicket-user@lists.sourceforge.net > Subject: Re: [Wicket-user] add ${label} as a validator > message variable > > Currently I override messageModel() (an approach I don't like > very much either, but until we have a better one). That is > part one. Part > 2: I modified AbstractValidator.error() to load a string resource with > validator.getClass() being the key (solving the default > problem) and finally I added a StringResourceLoader to > application settings which read string resources from > Application.properties which contain the ultimate default > values if none of the other files contain the properties. > > Juergen > > I add an additional StringResourceLoader to the settings which > > On 10/12/05, Igor Vaynberg <[EMAIL PROTECTED]> wrote: > > > The default message problem can be solved and it is working on my > > > laptop already. But IMO your request has two parts: > > > ${label} AND default messages. > > > And the ${label} part IMO is the one which needs some more > > > discussions. Any idea on how to solve the problem that a > context is > > > kind of anonymous where you prefer setLabel(). > > > What about a FormComponentContext which is derived from > HashMap and > > > which implements setLabel()? > > > > Imho this is not a two-part problem. How are you feeding > the label to > > the validators right now? > > > > -Igor > > > > > > > > > > Juergen > > > > > > On 10/12/05, Igor Vaynberg <[EMAIL PROTECTED]> wrote: > > > > > Lets continue the discussion on the list > > > > > > > > > > Currently I do not have the perfect solution. Your > > > approach is more > > > > > concise but I don't like that it > > > > > - is limited to label and tomorrow someone else will > ask for yet > > > > > another variable to add validators > > > > > > > > This discussion started out to find a solution to > provide default > > > > error messages. Label is a required variable for that. I > > > cant really > > > > think of any other variables that users may ask for, can > > > anyone give suggestions? > > > > > > > > > - the same will happen to formComponent. Add yet > something else > > > > > - though label and form components do have a > relationship, I'd > > > > > prefer not to add that "dependency" (see the previous > comments). > > > > > Label is not realy a part of a formComponent and inventing a > > > > > FormComponentLabel doesn't make it better, IMO. > > > > > > > > There is no dependency, you don't need to use > > > FormComponentLabel it is > > > > simply a convinience. > > > > > > > > > Just an idea: What about a "context" (HashMap) which > we'll add > > > > > to FormComponent instead of Label. Thus users are more > > > flexible and we > > > > > must not fear to add additional varaibles next month. The > > > "context" > > > > > would automatically be added to the validators > pre-existing model. > > > > > Instead of > > > > > > > > I like the context but I really don't like magic strings. > > > Externally I > > > > would still like to have set/getLabelModel(). Seems to me if > > > > people will want to use default error messages or provide their > > > own generic > > > > messages there will always need to be a label in the > context so we > > > > might as well make it easy/foolproof to set it. > > > > > > > > -Igor > > > > > > > > > > > > > > > > > > TextField tf=new TextField("hi"); > tf.setLabelModel(new Model("Hi > > > > > TextField")); > > > > > > > > > > you would do > > > > > TextField tf=new TextField("hi"); > > > > > tf.getValidatorContext().put("label", labelModel); > > > > > > > > > > The name implies it is related to validators (no magic) > > > and it is a > > > > > general solution. > > > > > What do you think? Are there any other idea how to solve > > > the problem > > > > > flexible and elegant? > > > > > > > > > > Juergen > > > > > > > > > > > > > > -------------------------------------------------------------------- > > > > > -- > > > > > > > > > > Initial Comment: > > > > > we need to be able to generate validator messages > with pattern > > > > > "${label} is a required field.". Somehow we need to > link a label > > > > > with a form component. > > > > > > > > > > > > > > > > > > > > > > > > -------------------------------------------------------------------- > > > > > -- > > > > > > > > > > >Comment By: Igor Vaynberg (ivaynberg) > > > > > Date: 2005-10-12 17:00 > > > > > > > > > > Message: > > > > > Logged In: YES > > > > > user_id=1293035 > > > > > > > > > > so what you are proposing, if i understand it correctly, > > > would work > > > > > like this: > > > > > > > > > > final IModel labelModel=new Model("Hi TextField"); > > > TextField tf=new > > > > > TextField("hi"); > > > > > > > > > > tf.add(new RequiredValidator() { messageModel() { return > > > > > super.messageModel().put("label", labelModel);}); > > > > > > > > > > tf.add(new EmailAddressTypeValidator() { > messageModel() { return > > > > > super.messageModel().put("label", labelModel);}); > > > > > > > > > > as opposed to: > > > > > > > > > > > > > > > TextField tf=new TextField("hi"); > tf.setLabelModel(new Model("Hi > > > > > TextField")); tf.add(RequiredValidator.getInstance()); > > > > > tf.add(EmailAddressTypeValidator.getInstance()); > > > > > > > > > > is that correct? > > > > > > > > > > > > > > > > > > > > > > > > -------------------------------------------------------------------- > > > > > -- > > > > > > > > > > Comment By: Juergen Donnerstag (jdonnerstag) > > > > > Date: 2005-10-12 16:35 > > > > > > > > > > Message: > > > > > Logged In: YES > > > > > user_id=544015 > > > > > > > > > > your are wrong. Validators already use a local model (see > > > > > AbstractValidator.messageModel) which are created while > > > > > executing > > > > > error() (that is, it is not a object variable). > > > > > Overriding messageModel does not at all prevent validator > > > singletons > > > > > etc. I suggest you take a look at AbstractValidator and > > > some of the > > > > > other Validators which already extend messameModel. > > > > > > > > > > Yes, most (not every) FormComponent have labels and and they > > > > > have forms and they have pages and they have ... Does it mean > > > we should > > > > > add this information as well. No. I still think overriding > > > > > messageModel is the better solution > > > > > > > > > > > > > > -------------------------------------------------------------------- > > > > > -- > > > > > > > > > > Comment By: Igor Vaynberg (ivaynberg) > > > > > Date: 2005-10-12 15:31 > > > > > > > > > > Message: > > > > > Logged In: YES > > > > > user_id=1293035 > > > > > > > > > > I dont see what value is added by adding the model into the > > > > > validators instead of the component. If we do it that way the > > > > > validators can no longer be singletons, also the same > > > model has to > > > > > be added to all validators for any form component > which wastes > > > > > space. FormComponent seems like the perfect place to do it. > > > > > > > > > > Most form components have label components "associated" with > > > > > them which have a model - we can create a formcomponentlabel > > > > > that will use the formcomponent's label model as its > own so we > > > > > will not be adding anything extra by having another > IModel ref > > > > > in the formcomponent. > > > > > > > > > > > > > > > > > > > -------------------------------------------------------------------- > > > > > -- > > > > > > > > > > Comment By: Juergen Donnerstag (jdonnerstag) > > > > > Date: 2005-10-12 06:10 > > > > > > > > > > Message: > > > > > Logged In: YES > > > > > user_id=544015 > > > > > > > > > > I'm -1 on adding a label model to FormComponent and > > > > > automatically use it for validators, because you actually can > > > > > extend the model used by the Validator by overriding > > > > > messageModel() already with any value you want, not > only a label > > > > > (only IntegerValidator's messageModel is final which must be > > > > > removed). > > > > > > > > > > The 2nd part of your request, Wicket to provide default > > > properties > > > > > files with properties like RequiredValidator=xx instead of > > > > > formname.input2.RequiredValidator to define default > > > values for all > > > > > RequiredValidators I agree upon with one addition. > Users must be > > > > > able to disable/enable it through apps.Settings. > > > > > > > > > > Juergen > > > > > > > > > > > > > > -------------------------------------------------------------------- > > > > > -- > > > > > > > > > > Comment By: Igor Vaynberg (ivaynberg) > > > > > Date: 2005-10-09 18:49 > > > > > - Show quoted text - > > > > > > > > > > Message: > > > > > Logged In: YES > > > > > user_id=1293035 > > > > > > > > > > no, this is not supported currently. > > > > > right now if i want to use a required validator i > need to do the > > > > > following: > > > > > page.properties: > > > > > formname.input1.RequiredValidator=Input 1 is required > > > > > formname.input2.RequiredValidator=Input 2 is required > > > > > > > > > > what i would like is to be able to do the following: > > > > > application.properties: > > > > > RequiredValidator=${label} is required and have the > > > > > formcomponent feed the proper ${label} value > > > > > > > > > > what this allows us is to provideis a list of > default validator > > > > > messages ie in webapplication.properties: > > > > > RequiredValidator=${label} is required > EmailTypeValidator=field > > > > > ${label} contans an invalid email address > > > > > > > > > > so the users dont have to provide > > > component*validator-count property > > > > > messages that are all mostly the same. > > > > > > > > > > i think the easiest way to accomplish this is to add an > > > IModel label > > > > > field to the FormComponent and have the abstract > > > validator set the > > > > > value of that model as the ${label} variable in the context. > > > > > > > > > > This would also allow as to have a nice > > > FormComponentLabel component > > > > > that uses the FormComponent's label model as its own model. > > > > > > > > > > > > > > > > > > > -------------------------------------------------------------------- > > > > > -- > > > > > > > > > > Comment By: Juergen Donnerstag (jdonnerstag) > > > > > Date: 2005-10-09 12:40 > > > > > > > > > > Message: > > > > > Logged In: YES > > > > > user_id=544015 > > > > > > > > > > I think this is already supported. Please see wicket src/test > > > > > wicket.util.tester.apps_1.CreateBook > > > > > (java/properties) and wicket-examples src/java > > > > > wicket.examples.forminput.FormInput (java/properties) > > > > > > > > > > > > > > > ------------------------------------------------------- > > > > > This SF.Net email is sponsored by: > > > > > Power Architecture Resource Center: Free content, downloads, > > > > > discussions, and more. > > > > > http://solutions.newsforge.com/ibmarch.tmpl > > > > > _______________________________________________ > > > > > Wicket-user mailing list > > > > > Wicket-user@lists.sourceforge.net > > > > > https://lists.sourceforge.net/lists/listinfo/wicket-user > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > ------------------------------------------------------- > > > > This SF.Net email is sponsored by: > > > > Power Architecture Resource Center: Free content, downloads, > > > > discussions, and more. > http://solutions.newsforge.com/ibmarch.tmpl > > > > _______________________________________________ > > > > Wicket-user mailing list > > > > Wicket-user@lists.sourceforge.net > > > > https://lists.sourceforge.net/lists/listinfo/wicket-user > > > > > > > > > > > > > ------------------------------------------------------- > > > This SF.Net email is sponsored by: > > > Power Architecture Resource Center: Free content, downloads, > > > discussions, and more. http://solutions.newsforge.com/ibmarch.tmpl > > > _______________________________________________ > > > Wicket-user mailing list > > > Wicket-user@lists.sourceforge.net > > > https://lists.sourceforge.net/lists/listinfo/wicket-user > > > > > > > > > > > > > > > > > > > ------------------------------------------------------- > > This SF.Net email is sponsored by: > > Power Architecture Resource Center: Free content, downloads, > > discussions, and more. http://solutions.newsforge.com/ibmarch.tmpl > > _______________________________________________ > > Wicket-user mailing list > > Wicket-user@lists.sourceforge.net > > https://lists.sourceforge.net/lists/listinfo/wicket-user > > > > > ------------------------------------------------------- > This SF.Net email is sponsored by: > Power Architecture Resource Center: Free content, downloads, > discussions, and more. http://solutions.newsforge.com/ibmarch.tmpl > _______________________________________________ > Wicket-user mailing list > Wicket-user@lists.sourceforge.net > https://lists.sourceforge.net/lists/listinfo/wicket-user > > > ------------------------------------------------------- This SF.Net email is sponsored by: Power Architecture Resource Center: Free content, downloads, discussions, and more. http://solutions.newsforge.com/ibmarch.tmpl _______________________________________________ Wicket-user mailing list Wicket-user@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/wicket-user
