Dead keys followed by space

[Johan Corveleyn]

Thanks, and thank you Martin for filing the issue
https://bugs.openjdk.org/browse/JDK-8340982. I'll try to create an account
on openjdk.org to be able to watch the issue :-).

Tangentially:
For some reason mails on the openjfx-dev mailinglist coming from
martin at martinfox.com are not coming through to me. I only saw this thanks
to Thiago's reply below (which prompted me to realize that I have missed
Martin's mail). It's in the archives on
https://mail.openjdk.org/pipermail/openjfx-dev/2024-September/thread.html
alright, which is where I saw Martin's latest reply about filing the issue.
It seems mails from mfox at openjdk.org are coming through to me, but from
martin at martinfox.com not. Maybe some overzealous filtering by Gmail (not
sure whether it's only an issue for me, or also for other gmail users).
I'll try to investigate a bit more why gmail seems to be dropping these
mails for me :-/.

-- 
Johan

On Wed, Sep 25, 2024 at 6:38 PM Thiago Milczarek Sayão <
thiago.sayao at gmail.com> wrote:

> Johan,
>
> Trust Martin, as I know nothing about Windows :)
> Sorry for misleading you on the IME subject.
>
> Em qua., 25 de set. de 2024 às 13:24, Martin Fox <martin at martinfox.com>
> escreveu:
>
>> Johan,
>>
>> Thanks for providing all the details. It does sound like a bug. I will
>> take a look.
>>
>> Since this is Windows altering the IME setup probably won’t have any
>> effect. On Mac and Linux dead keys are delivered through the IME pathway
>> and are presented to JavaFX as InputMethod events. On Windows dead keys
>> don’t go through this pathway and are presented as KeyEvents.
>>
>> Martin
>>
>> On Sep 25, 2024, at 8:21 AM, Kevin Rushforth <kevin.rushforth at oracle.com>
>> wrote:
>>
>> I would not recommend internal APIs, unless someone wants to build JavaFX
>> to help diagnose it.
>>
>> In any case it sounds like Johan has discovered a bug -- at least on
>> Windows -- in the key handler for dead keys. We should test this on macOS
>> and Linux as well.
>>
>>  -- Kevin
>>
>> On 9/25/2024 8:14 AM, Thiago Milczarek Sayão wrote:
>>
>> Johan,
>>
>> It's an internal API, but we can use it for testing purposes (I'm not
>> sure if there's a public API for that).
>> SceneHelper.enableInputMethodEvents(scene, false);
>> IME is used for entering complex characters in some languages such as
>> Chinese, Korean, Japanese, etc.
>>
>> On Linux, setting the keyboard to en_US will also add the extra space,
>> except if I choose the en_US with dead keys variation.
>>
>> If you try it on notepad, is it the same behaviour? If not, I think it
>> might be a bug.
>>
>> -- Thiago.
>>
>>
>> Em qua., 25 de set. de 2024 às 11:12, Johan Corveleyn <jcorvel at gmail.com>
>> escreveu:
>>
>>> Hi Thiago,
>>>
>>> Thank you for your answer. I am experiencing this issue on Windows
>>> (Windows 10 and 11).
>>>
>>> Now, it dawns on me that:
>>> - I have a QWERTY keyboard (standard US layout).
>>> - I have my keyboard layout setting in Windows configured to "US
>>> International". That way the keys for ~, ^, ', " become "dead keys"
>>> (waiting for next keystroke).
>>> - If I change my keyboard layout setting to "US" then I can't
>>> reproduce anymore because those keystrokes no longer act like dead
>>> keys (~, ^, ... immediately give that character).
>>>
>>> I don't know how this is handled on Linux or Mac. Do they also have
>>> something like a "US International" keyboard layout setting so ^
>>> becomes a dead key? Or does it work differently if I want to type â or
>>> ë, or a standalone ^?
>>>
>>> I'm not sure what you mean by disabling IME or how to know that it's
>>> enabled. Can I enable/disable this programmatically in JavaFX? Or is
>>> this an OS keyboard layout setting (anyway: just having a standard "US
>>> input" makes this non reproducible because I can no longer type dead
>>> keys)?
>>>
>>> If you have a keyboard input with dead keys it's super easy to test
>>> this with a simple program like:
>>> [[[
>>> public class DeadKeysFX extends Application {
>>>     @Override
>>>     public void start(Stage stage) {
>>>         TextField textField = new TextField();
>>>         Scene scene = new Scene(textField);
>>>         stage.setScene(scene);
>>>         stage.show();
>>>     }
>>> }
>>> ]]]
>>>
>>> Just type ^+<space> and see whether you get one or two characters. In
>>> all non-JavaFX programs (including Java Swing) I get a single '^'. In
>>> JavaFX I get a '^ '.
>>>
>>> --
>>> Johan
>>>
>>> On Tue, Sep 24, 2024 at 5:16 PM Thiago Milczarek Sayão
>>> <thiago.sayao at gmail.com> wrote:
>>> >
>>> > Hi Johan,
>>> >
>>> > It might vary be platform. Which one are you using? (Windows, Mac,
>>> Linux).
>>> >
>>> > Try disabling IME (Input Method Editor) and see it it works.
>>> >
>>> > - Thiago
>>> >
>>> > Em ter., 24 de set. de 2024 11:51, Johan Corveleyn <jcorvel at gmail.com>
>>> escreveu:
>>> >>
>>> >> Hi,
>>> >>
>>> >> (This is my first post here, hope I'm following the right path)
>>> >>
>>> >> It seems JavaFX TextFields (and friends) do not automatically convert
>>> >> <dead key>+<space> into simply <dead key character>. They make it into
>>> >> <dead key character>+<space>, which is very atypical. I would consider
>>> >> this behavior a bug, since it is different from any editor I know (and
>>> >> makes it very hard to enter a dead key character on its own). We ran
>>> >> into this with Java 8, I also reproduced it with openjdk 21 + openjfx
>>> >> 23.
>>> >>
>>> >> For instance if in a JavaFX TextField I type a '^' keystroke, it waits
>>> >> for the next keystroke (which is normal since it's a dead key,
>>> >> possibly followed by a character on which to put the '^'). But if I
>>> >> then type <space> I expect a simple '^' to appear. Instead, in JavaFX,
>>> >> '^ ' appears. This does not happen in Swing, nor in any editor or
>>> >> shell or ...
>>> >>
>>> >> Background context: a user of our JavaFX application couldn't
>>> >> authenticate with their password (typed in a PasswordField). After an
>>> >> awful lot of troubleshooting we found that they used a '^' in their
>>> >> password. Of course the user didn't notice that after typing ^+space
>>> >> two dots appeared in the PasswordField. Now that user knows they have
>>> >> to backspace after typing ^+space ...
>>> >>
>>> >> I suppose inserting a Swing JPasswordField in our JavaFX app would
>>> >> work around this issue, but ... isn't there a better solution?
>>> >> Shouldn't this be regarded as a bug?
>>> >>
>>> >> --
>>> >> Johan
>>>
>>
>>
>>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://mail.openjdk.org/pipermail/openjfx-dev/attachments/20240926/803982a4/attachment-0001.htm>