Proposal: Focus Traversal API

[Andy Goryachev]

I like "try" very much, does not clash with multiple requestXXX in autocompletion.

tryMoveFocus(TraversalDirection.UP);

-andy



From: openjfx-dev <openjfx-dev-retn at openjdk.org> on behalf of Nir Lisker <nlisker at gmail.com>
Date: Wednesday, October 23, 2024 at 11:54
To: Kevin Rushforth <kevin.rushforth at oracle.com>
Cc: openjfx-dev at openjdk.org <openjfx-dev at openjdk.org>
Subject: Re: Proposal: Focus Traversal API
My only problem with a more assertive “moveFocus” is that it implies a success

True, but the JDK already uses it everywhere. Mutation methods on collections are all named as definitive actions even if they don't succeed. In general, setters might also fail (usually in the case of NPE), but they are named `setXxx`.
I would say it depends on what failure looks like. If you throw an exception, then this name is fine because you're going to be told it failed. If you return a boolean to indicate success (a design I don't like, but collections do it), it might also be fine. However, if you "fail silently", it could be a problem.
In my code, I do sometimes use the `try` prefix: `tryToFocus` in this case. Then again, I prefer prose over convention so my method names can be rather... descriptive. The advantage I find with writing descriptive names is that they get clunky if the method does too much, which prompts me to split the method, resulting in what I would call cleaner code. Not everyone agrees on this.
Another way to name a method is to use the argument name as part of its name. E.g., instead of `moveFocus(Node node)`, you would write `moveFocusTo(Node node)`. Here, the argument is a glorified direction, so it partially works: `moveFocusTo(TraversalDirection.NEXT)` works well, `moveFocusTo(TraversalDirection.UP)` is OK.

On Wed, Oct 23, 2024 at 9:15 PM Kevin Rushforth <kevin.rushforth at oracle.com<mailto:kevin.rushforth at oracle.com>> wrote:
I think we just need a good name. I agree that "predict" isn't the API we want.

How about "requestFocusTraversal"? It has symmetry with requestFocus.

-- Kevin

On 10/23/2024 11:05 AM, Andy Goryachev wrote:
I don't think prediction should be the functionality we want in the Node.  This functionality, in my opinion, belongs to the traversal policy.

All we want is to navigate away from the specified node.

It might have been easier had we kept the original FocusTraversal class - not only one can add a bunch of convenience methods (predict, infer, contemplate, etc.) but we could have added applicationFocusOwnerProperty() etc.

Do we need another brainstorming round?

-andy




From: openjfx-dev <openjfx-dev-retn at openjdk.org><mailto:openjfx-dev-retn at openjdk.org> on behalf of Michael Strauß <michaelstrau2 at gmail.com><mailto:michaelstrau2 at gmail.com>
Date: Wednesday, October 23, 2024 at 10:58
To:
Cc: openjfx-dev at openjdk.org<mailto:openjfx-dev at openjdk.org> <openjfx-dev at openjdk.org><mailto:openjfx-dev at openjdk.org>
Subject: Re: Proposal: Focus Traversal API
How about „predictFocus“, which will return a Node without changing focus?

This would give applications even more flexibility, because they can then decide if they want to go ahead and actually change focus by calling predictFocus()?.requestFocus()


Andy Goryachev <andy.goryachev at oracle.com<mailto:andy.goryachev at oracle.com>> schrieb am Mi. 23. Okt. 2024 um 19:16:
Thank you for suggestion!

My only problem with a more assertive “moveFocus” is that it implies a success, while this API may or may not succeed (similarly to requestFocus).

The word "shift" in "shiftFocus" might clash with the "shift" key, don't really like that.

Maybe "requestTraversal" although it will expand auto-completion list that has 2 items already.  Welcome any alternative, as long as it's not creat().

-andy


From: openjfx-dev <openjfx-dev-retn at openjdk.org<mailto:openjfx-dev-retn at openjdk.org>> on behalf of Michael Strauß <michaelstrau2 at gmail.com<mailto:michaelstrau2 at gmail.com>>
Date: Wednesday, October 23, 2024 at 10:05
To:
Cc: openjfx-dev at openjdk.org<mailto:openjfx-dev at openjdk.org> <openjfx-dev at openjdk.org<mailto:openjfx-dev at openjdk.org>>
Subject: Re: Proposal: Focus Traversal API
Maybe we can polish the naming a bit:

When used as a transitive verb, "traverse" means (according to
Merriam-Webster): to go or travel across or over, to move or pass
along or through.

MW gives the example: "light rays traversing a crystal". So one thing
is moving (light rays), while the other thing is static (a crystal).
Cambridge gives the example "Stanley traversed the continent from west
to east." Again, "Stanley" is moving, "the continent" is static.

Since "traverseFocus" is a transitive verb followed by a direct noun,
it sounds to me like "focus" is the thing that can be traversed over
or across.
As an analogy, let's call the method "traverseContinent" for a moment.
It becomes obvious that "continent" is the space which is traveled
across, and not the thing travelling.

In our case, "focus" is the moving thing, not the thing that it is
moved across. I would use something like "moveFocus", "shiftFocus", or
something along those lines.

As for "TraversalDirection", I think it is acceptable, but I'll note
that to me, "direction" implies one of the four directional modes, not
the two logical modes (previous and next).

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://mail.openjdk.org/pipermail/openjfx-dev/attachments/20241023/907c04d2/attachment-0001.htm>