Fix#289: Validate subset name for database

[davidlatwe]

This PR resolve #289, Creator will ensure the subset name is valid for MongoDB.

What's changed?

• Validate subset name in Creator, and the GUI will indicate the validation result

[davidlatwe]

Demo

[tokejepsen]

Maybe this is helping the users too much, but maybe the "Invalid subset name..." message could tell the users which characters are invalid?

[davidlatwe]

Yeah, I thought about that, too. But seems complex to implement 😟

[tokejepsen]

Hmm since the "Create" button gets invalidated, it might not be a must have. Great work!

[davidlatwe]

Haha, turns out it's not that hard :P
Implemented in commit 79ed093, and here's the demo (Note that the space char will be quoted)

What do you think ?

[tokejepsen]

Amazing!

[davidlatwe]

Wait, it seems that QLineEdit already has a feature to validate the input text, QLineEdit.setValidator.
Well try this when I get back from 2 days vacation.

[davidlatwe]

I have change to use QValidator to block invalid inputs, also add a tool tip.
Here the result :

Note that the invalid characters will not be accepted at all, so I add a border color transition animation for better feedback.

Please let me know what you think about this change :)

[tokejepsen]

Looks great!

[BigRoy]

From an artist/interface perspective this looks good to me. Code-wise I only had some questions, because it seems quite an amount of lines of code for the visual result. It looks good though.

[BigRoy]

To be sure, the only reason we require to define our own class is so we can report to the user which characters were invalidated in particular. Correct? If so, then this class is fine. Otherwise, this seems redundant.

[davidlatwe]

Correct?

Not entirely, the main reason I subclass the QtGui.QRegExpValidator was to re-implement the function validate, so that I could emit a signal when the input has invalid character. I could not find any built-in way to implement this notification on QLineEdit, it just works silently.

Reporting which characters exactly were invalid was something I picked up on the way :)

[BigRoy]

Ah. If you only wanted to have the invalid signal, it's QLineEdit.inputRejected but it won't signal what characters were invalid I believe + it's only been introduced since Qt 5.12 so I guess we can't rely on it.

I guess currently subclassing like this really is the only way.

[BigRoy]

Not sure if there's enough reason to abbrevaite _color_status to _cstatus. I don't mind it. Just wondering.

[davidlatwe]

Hmm, this should be improved for reading.

[mottosso]

I love it!

[davidlatwe]

Well merge this tomorrow !