[0.7-M4] Inserting text via user keyboard input inserts text in reverse order because caret position doesn't advance forward after input

[ghost]

Replicate

1. Run the Java Keywords demo
2. Start typing

Expected Results
The caret advances with each character typed.

Actual Results
The caret retains its position, which causes the typed text to be entered backwards.

[JordanMartinez]

I'm guessing you're using the new 0.7-M4 release? That's probably a bug from #468. :-/

[JordanMartinez]

Nope. I was wrong. The bug start showing up in #463. After I moved the model-API to ESD, every time one types a letter, it moves the caret to the right twice so that the caret is one character farther than it should be. After I moved the view API to the area, the behavior reverted to what @DaveJarvis just now described in this issue.

[ghost]

Correct, this is in the 0.7-M4 release. 0.7-M3 worked as expected.

[JordanMartinez]

I've updated the release notes to note this issue.

[JordanMartinez]

@TomasMikula Would it be possible to re-release the 0.7-M4 release using the updated master branch that accounts for this issue? Or would this require another release?

[JordanMartinez]

Just fixed it. Dave, could you try it out to insure there are no other regressions? Although I already tagged you in the issue, is it possible, Tomas, to re-release the 0.7-M4 release with the current master branch? Or does this require a new release? Blessings, Jordan

…

On Mon, Mar 27, 2017 at 11:50 PM, Dave Jarvis ***@***.***> wrote: Correct, this is in the 0.7-M4 release. 0.7-M3 worked as expected. — You are receiving this because you commented. Reply to this email directly, view it on GitHub <#472 (comment)>, or mute the thread <https://github.com/notifications/unsubscribe-auth/AIBfbV8rBsp3VuyunKz6yXKXVdXL6glwks5rqK2tgaJpZM4MrFxc> .

[ghost]

@JordanMartinez, once the new release is available, I'll try it out.

I'd suggest making a new release, 0.7-M5. (Mostly to avoid caching issues for anyone who already has 0.7-M4 downloaded.)

[TomasMikula]

I pushed a new SNAPSHOT release, can you try that out?

I think it used to be possible to override a release, but yeah, there can be caching issues.

[JordanMartinez]

:-/ Dang...

[ghost]

SNAPSHOT release has resolved this issue.

[JordanMartinez]

Can a new milestone release be made then?

@TomasMikula Besides the bugs and dependency bugs tagged in this project's list of issues, is there anything else that needs to be done before a stable 0.7 release can be made?

[JordanMartinez]

It is possible to fix this regression in the 0.7-M4 release by overriding the replace method like so (Updated to account for TextChange's exception in some situations):

public void replace(int from, int to, StyledDocument<PS, SEG, S> document) {
    if (from == to && document.length() == 0) {
        return; // ignore changes that do nothing. Or TextChange throws IllegalArgumentException
    }
    super.replace(from, to, document);
    int newCaretPos = getCaretPosition() + document.length();
    selectRange(newCaretPos, newCaretPos);
}

[TomasMikula]

Released 0.7-M5. As for 0.7, I don't know, mainly be it more stable, API-wise and bug-wise.

[JordanMartinez]

Ok. Bug-wise, there's 2 RTFX-specific bugs that can be fixed somewhat quickly (#423 & #362). I'm not sure what is causing the problem with #265 and I can't test (and thus fix) it because I don't use/have a Mac. Then there's the Flowless-related bugs (#412, #390, and probably #260).

How could the API be made more stable?