What is a generally accepted way to validate preference values in a Firefox extension, specifically when using the prefwindow mechanism in XUL?
I am introducing some new preferences in one of my extensions that I would like to validate before the preferences window is closed. If there's an error, the user should be allowed to correct the issue, and then proceed. I see that the prefwindow element has two potentially useful functions to help in this regard:
onbeforeaccept
ondialogaccept
The former seems to have an associated bug (Bug 474527) that prevents the prefwindow from remaining open when returning false from that function. This is bad in that it doesn't give the user an opportunity to immediately correct their mistake.
The latter appears to have the problem that the preferences get saved prior to exiting, which leaves the preferences in a bad state internally.
In addition, the prefwindow mechanism supports the browser.preferences.instantApply option, in which preference values are written immediately upon updating the associated control. This makes validation extra tricky. Is there a clean way to validate custom preferences in a Firefox extension, allowing the user to correct any potential mistakes?
Normally you would want to validate the preferences when they are changed. That's something that onchange attribute (and the corresponding change event) is good for:
<preference name="preference.name" onchange="validate(this);"/>
The event is fired after the preference value changes. There are two drawbacks:
In case of instantApply the new preference value is already saved, too late to validate and decline.
For text fields the preferences are saved every time a new character is typed. This becomes ugly if you report validation failure while the user is still typing.
You can solve the first issue by intercepting the change events for the actual input fields. For example, for a text field you would do:
<input preference="preference.name"
oninput="if (!validate(this)) event.stopPropagation();"
onchange="if (!validate(this)) { event.stopPropagation(); this.focus(); }"/>
So changes that don't validate correctly don't bubble up to the <prefpane> element and don't get saved. The events to listen to are: input and change for text fields, command for buttons and checkboxes, select for the <colorpicker> element.
The second issue is tricky. You still want to validate the input when it happens, showing the message immediately would be bad UI however. I think that the best solution is to assume for each input field initially that it is still "in progress". You would only set a flag that the value is complete when you first see a blur event on the field. That's when you can show a validation message if necessary (ideally red text showing up in your preference page, not a modal prompt).
So to indicate what the final solution might look like (untested code but I used something like that in the past):
<description id="error" hidden="true">Invalid preference value</description>
<input preference="preference.name"
_errorText="error"
onblur="validate(event);"
oninput="validate(event);"
onchange="validate(event);/>
<script>
function validate(event)
{
// Perform actual validation
var field = event.target;
var valid = isValid(field);
// If this is the blur event then the element is no longer "in progress"
if (event.type == "blur")
{
field._inputDone = true;
if (!valid)
field.focus();
}
// Prevent preferences changing to invalid value
if (!valid)
event.stopPropagation();
// Show or hide error text
var errorText = document.getElementById(field.getAttribute("_errorText"));
errorText.hidden = valid || !field._inputDone;
}
</script>
If you want to validate values as soon as the field is changed so you can handle the instantApply case, you could hook into the change events for the individual fields (e.g. oninput for a textbox). Display an error message and force the focus back to the field if the value is invalid. You can either set it back to a valid value automatically or block the user from closing the dialog until it is fixed.
Related
From https://docs.ckeditor.com/ckeditor4/docs/#!/api/CKEDITOR.editor-event-change
If it is important not to get the change event fired too often, you should compare the previous and the current editor content inside the event listener. It is not recommended to do that on every change event.
I don't understand the above ... is there possibly a typo?
Is it trying to say that a call such as checkDirty on every change event could result in poor performance?
Would using an interval timer be a good approach to limit the number of checkDirty calls made.
For example, do not call checkDirty unless at least 1 second had passed from its last invocation?
The use case I have is enabling or disabling a save button based on the editor content differing from its last saved state.
I deal with this situation by always having the save button enabled and:
1) Every time a user presses the save button, I alert him that the contents have been saved and I clear the dirty flag:
CKEDITOR.instances.editor1.resetDirty();
2) I register an onbeforeunload event handler to alert him if he tries to exit the webpage having unsaved data:
window.onbeforeunload = function() {
if (CKEDITOR.instances.editor1.checkDirty()) {
return 'You will lose the changes made in the editor if you don't save them.';
}
}
https://react.semantic-ui.com/modules/search
Below is images of how the semantic react ui search widget looks like. In the second image, I was wondering how you can put a prompt message to indicate to the user a message on what the search bar is meant for. In this case, it's "search". When the user types in, it erases the Search and starts reflecting what the user is typing. I thought it would be defaultValue, but it seems that you can't have value and defaultValue set at the same time. I still want to be able to read what the set value is when the user types into the box.
Thanks,
Derek
You can use defaultValue as initial value in component, no problem.
Then read the user input value in event (onBlur for instance) like this:
onBlur(e) {
e.preventDefault();
console.log(e.target.name, e.target.value);
}
If you want to read value each new character pressed you can use in onSearchChange event:
onSearchChange(e) {
e.preventDefault();
console.log(e.target.name, e.target.value);
}
EDIT: included accepted answer in comment below:
Worked:
placeholder={"text"}
for Semantic React UI Search
I've been trying to implement what's been asked in this Stack Overflow question, here:
Calculation for status in Archer GRC based on date
Trying to create a status field based on a number of Value Lists that
users select from, but a request has been made that we check a date
field for a value to ensure an estimated date has been set so that the
calculation can determine if the status of the record is "In
Progress", "Late" or "Not Started".
...and now, I have a requirement for an actual popup warning message of some sort to prompt the user to make sure the date field is not blank.
How would I add this functionality?
In order to deliver the functionality you are looking for you have to use a "Custom Object". It is an object you put on the layout of the application in Archer that contains JavaScript code. This code will be executed as soon as the form of the application is loaded. There is a special type of the field "Custom Object" available in the Layout editor for each application in the Application Builder in Archer.
Note - I don't recommend to use custom objects in general and neither RSA Support. Every time you modify the layout in the given application, you have retest and sometimes correct IDs for your custom object. You can write an ID independent custom object and use field names, but in this case custom object will have more code. I prefer to make custom objects as short as possible.
Your custom object should do the following:
Override the behavior of the "Save" and "Apply" button in the top tool bar available for every application form in Archer.
Once "Save" and "Apply" buttons are "overwritten", every time they are clicked on your function will be called. So you need to create a click handler function.
Your click handler function will check values user is required to populate and will either return warning, or will call the original handler for "Save/Apply" buttons.
This is a code template you can start with:
<script type="text/javascript">
// ids are used to locate buttons
var buttons_ids = [
"master_btnSave", // "Save" button ID
"master_btnApply" // "Apply" button ID
];
// parameters are used in the "onclick" default handlers to call original handlers
var buttons_parameters = [
"master$btnSave", // "Save" parameter
"master$btnApply" // "Apply" parameter
];
document.getElementById(buttons_ids[0]).onclick = function(){ Validator_of_required_fields(buttons_parameters[0])};
document.getElementById(buttons_ids[1]).onclick = function(){ Validator_of_required_fields(buttons_parameters[1])};
// end of the script body
//==== Validator function attached to Save and Apply buttons
function Validator_of_required_fields(parameter){
// ids of the input fields to validate
var inputs_to_validate_ip_address = [ "master_DefaultContent_rts_XXX_YYY_t" ];
// jQuery selector is used here. Archer v5.x has jQuery library loaded by default
// you will need to modify this selector
var field_value = $('#'+inputs_to_validate_ip_address[0]+':first').val();
if(field_value.length = 0) {
// Here you are calling Archer Warning function
var msg = "[Text to display to user]";
var title = 'Required Field';
WarningAlert(msg,title);
return false;
};
// default onclick processor
ShowAnimationAndPostback(parameter);
return false;
};
Some comments on this code:
You will need to modify the validation function to work with values stored in the fields you need.
I used a rather 'unusual' way to override the behavior of the "Save" and "Apply" buttons using the following code:
document.getElementById(buttons_ids[0]).onclick = function(){ bla, bla, bla }There are simpler way to do the same, but this way custom object works fine in IE8-11, FF, Chrome and Opera. Let me know if you find a simpler way to override buttons that is browser agnostic.
Function WarningAlert(msg,title); is a build-in Archer warning message function. It worked fine in Archer v5.4. You might need to use simple JavaScript Alert function if WarningAlert doesn't work in your version of Archer.
Note that behavior of the "Save" and "Apply" buttons might be overwritten back to default in case if user opens up any pop-up dialog windows to populate a value list or cross-reference field. If that is the case, you will have to wrap the code provided into another function and attach it to the OnLoadWindow event (or similar).
I try to avoid using any JavaScript libraries in my custom objects. This way it is simpler to support them and you have less dependencies. I used jQuery in the provided example only because Archer already uses this library once the page is loaded.
Flak, make sure to test your custom object very well and good luck!
I am new to ReactiveCocoa, but I think it's very nice and outstanding technique for reducing code complexity. I just started experiencing with the framework, and not everything is clear for me at the moment, so excuse me if my problem can be solved in some obvious way.
In my app I have login view controller with simple form contains two text fields (username and password) and a button. I would like the button to be disabled if any of two text fields is empty. So, I wrote this code:
RAC(self.loginButton, enabled) =
[RACSignal combineLatest:#[self.userTextField.rac_textSignal,
self.passwordTextField.rac_textSignal]
reduce:^(NSString *username,
NSString *password) {
BOOL valid = (username.length > 0 && password.length > 0);
return #(valid);
}];
It's very simple and it's working. The problem is that one of my text fields (the password field) has secureTextEntry and clearsOnBeginEditing properties set to YES. I will try to explain unwanted behavior that I am experiencing with this configuration:
Let's assume that both username and password fields are NOT empty. In this case the button is enabled. When user taps on password field, it becomes first responder (keyboard appears and user can enter his password), but because of clearsOnBeginEditing being set to YES for that field, the previously entered password is cleared from the text field. That's way password field is now empty. The problem is that signal is not being sent, so the button remains enabled, despite the password field is empty.
My first idea to solve this issue (well, more like workaround solution) was to observe isFirstResponder property on password field beside observing text changes. That's way the block that checks if button should be enabled would be called when password field becomes first responder. I don't know if this solution works, because I have no idea how to implement it using ReactiveCocoa. I have looking for creating a signal for isFirstResponder property changes, but without a luck. It might be not the best approach in order to solve this issue, but nothing comes to my mind at this point.
Then, the question is: how to observe isFirstResponder property with ReactiveCocoa?
And more general question: how to observe text field's text changes when clearsOnBeginEditing is set to YES?
UPDATE:
I found out that I can create signal for UIControlEventEditingDidBegin event that should give me substitution of observing isFirstResponder property changes:
[self.passwordTextField rac_signalForControlEvents:UIControlEventEditingDidBegin]
Unfortunately this does not solve the issue. Now I understand that field is cleared AFTER it becomes first responder, and clearing field automatically after it becomes first responder does not send signal for text changes. That's way when validation block is executed it still thinks that password field is not empty, and the button remains enabled despite password field was cleared and it's empty.
Unfortunately the -rac_textSignal only listens for UIControlEventEditingChanged. If UIControlEventEditingDidBegin were added, you'd be all set.
I suppose you could patch this into it and submit a pull request?
- (RACSignal *)rac_textSignal {
#weakify(self);
return [[[[[RACSignal
defer:^{
#strongify(self);
return [RACSignal return:self];
}]
concat:[self rac_signalForControlEvents:UIControlEventEditingChanged|UIControlEventEditingDidBegin]]
map:^(UITextField *x) {
return x.text;
}]
takeUntil:self.rac_willDeallocSignal]
setNameWithFormat:#"%# -rac_textSignal", [self rac_description]];
}
In my little cocoa application I have bound the properties of a class to some text fields with help of a NSObjectController. The only problem I have so far: you always have to leave a text field before the NSObjectController updates the class with the current input.
This becomes a problem if the user doesn't leave a texfield and clicks on a Save/Submit Button right away. The class doesn't contain the current input. Always a bad thing.
I am looking for a way to avoid this. Like telling the NSObjectController to get the current input even if the user had exited the field. If this is possible I could put this command in the save-Method before saving and all would be fine.
Send a commitEditing message to your controller in the handler for the OK button. This will do what you're asking for. It's as simple as:
- (void)save:sender {
if (![self.myObjectController commitEditing]) {
// Handle error when object controller can't commit editing
}
// Other stuff
}
If you go to the text field's value binding and check the "Continuously Updates Value" option, that will cause the new value to be set on the model object each time the user changes it, i.e. once for each keystroke. That would ensure that the model had the correct value before closing the window, though it may be a bit overkill, depending on what the effects (if any) are of the value being set in your data model.