1799578 - [CTW] Text Cursor Indicator misplaced on multiline textareas with padding

Status: RESOLVED FIXED

(Core :: Disability Access APIs, defect, P3)

Description

[Asa Dotzler [:asa]]

Attachment: text-cursor-indicator-bug.png

In bug 1797222, we fixed a problem with the Text Cursor Indicator refusing to track text across wraps or line breaks -- or something like that :-)

Now it does track rows of text better, but in textareas with padding set, the indicator ends up below the current line (and perhaps a few pixels too far to the right.) I'll attach a screenshot that makes it clear what I'm describing.

Another note, and maybe it deserves a separate bug, but if I scroll the document, the Text Cursor Indicator stays in the same place rather than staying "attached" to the actual cursor.

Comment 1

[James Teh [:Jamie] (away returning 28 Jan)]

Test case:
data:text/html,<textarea style="padding: 10px;">ab%0acd
I guess our y coord for continuations is off somehow in the presence of padding, but I don't have any clue why.

Comment 2

[James Teh [:Jamie] (away returning 28 Jan)]

(In reply to Asa Dotzler [:asa] from comment #0)

Another note, and maybe it deserves a separate bug, but if I scroll the document, the Text Cursor Indicator stays in the same place rather than staying "attached" to the actual cursor.

I think it'd be good to have a different bug for this. Also, does this one occur with CTW disabled too or only CTW enabled?

Comment 3

[Morgan Reschenberg [:morgan]]

(In reply to James Teh [:Jamie] from comment #2)

(In reply to Asa Dotzler [:asa] from comment #0)

Another note, and maybe it deserves a separate bug, but if I scroll the document, the Text Cursor Indicator stays in the same place rather than staying "attached" to the actual cursor.

I think it'd be good to have a different bug for this. Also, does this one occur with CTW disabled too or only CTW enabled?

I've got a mochitest for the padding issue, which passes with the cache off. So technically a regression. Not sure about the scrolling

Comment 4

[Morgan Reschenberg [:morgan]]

Attachment: Bug 1799578: Compute relative offsets for continuations to avoid double counting padding when computing CharBounds r?Jamie

Comment 5

[Pulsebot]

Pushed by [e-mail]:
https://hg.mozilla.org/integration/autoland/rev/617bbc5ca072
Compute relative offsets for continuations to avoid double counting padding when computing CharBounds r=Jamie

Comment 6

[Serban Stanca [:SerbanS]]

https://hg.mozilla.org/mozilla-central/rev/617bbc5ca072