UIA in conhost: incorrect word boundaries when text is selected #15950

Open
opened 2026-01-31 04:53:11 +00:00 by claunia · 10 comments
Owner

Originally created by @codeofdusk on GitHub (Nov 23, 2021).

Windows Terminal version

Version of OpenConsole included with the wt 1.12 release.

Windows build number

10.0.22000.348

Other Software

NVDA master (2022.1) with UIA.

Steps to reproduce

Paste the following text into a conhost. On my system, this line wraps at the word "upgrade":

tree.IncompatibleFormatError: This file is too new for the currently running version of Treemendous (0.0.0). Please upgrade to Treemendous 42.0.0 or later.

Then:

  • Select a word, such as by moving to the top of the console and searching for the word "please" with CTRL+f.
  • navigate forward in the window by word.

Expected Behavior

word boundaries are correctly calculated: after the wrapping, NVDA says "upgrade", "to", "Treemendous", etc.

Actual Behavior

NVDA reads from the start of the line up to some arbitrary point, can't tell what's going on???

Originally created by @codeofdusk on GitHub (Nov 23, 2021). ### Windows Terminal version Version of OpenConsole included with the wt 1.12 release. ### Windows build number 10.0.22000.348 ### Other Software NVDA master (2022.1) with UIA. ### Steps to reproduce Paste the following text into a conhost. On my system, this line wraps at the word "upgrade": `tree.IncompatibleFormatError: This file is too new for the currently running version of Treemendous (0.0.0). Please upgrade to Treemendous 42.0.0 or later.` Then: * Select a word, such as by moving to the top of the console and searching for the word "please" with CTRL+f. * navigate forward in the window by word. ### Expected Behavior word boundaries are correctly calculated: after the wrapping, NVDA says "upgrade", "to", "Treemendous", etc. ### Actual Behavior NVDA reads from the start of the line up to some arbitrary point, can't tell what's going on???
Author
Owner

@codeofdusk commented on GitHub (Nov 23, 2021):

This does not repro in wt.

CC @carlos-zamora, @DHowett.

@codeofdusk commented on GitHub (Nov 23, 2021): This *does not* repro in wt. CC @carlos-zamora, @DHowett.
Author
Owner

@codeofdusk commented on GitHub (Dec 5, 2021):

Confirmed repro of this issue in FORMATTED conhost, but looks like this is the only remaining word boundary bug!

@codeofdusk commented on GitHub (Dec 5, 2021): Confirmed repro of this issue in `FORMATTED` conhost, but looks like this is the only remaining word boundary bug!
Author
Owner

@codeofdusk commented on GitHub (Dec 5, 2021):

On further testing, it is important that the selected text must be on a line that wraps, and the selection must be somewhere on that line. NVDA+f9/f10 also works for selection here (not just CTRL+f), so I presume the mouse would as well.

@codeofdusk commented on GitHub (Dec 5, 2021): On further testing, it is important that the selected text *must* be on a line that wraps, and the selection must be somewhere on that line. NVDA+f9/f10 also works for selection here (not just CTRL+f), so I presume the mouse would as well.
Author
Owner

@zadjii-msft commented on GitHub (Dec 6, 2021):

This might be related to #5800, or some of the weird line wrapping bugs that I've at least been linking up to that one. I'm thinking particularly the #6901, #10833, #11739 subset.

@zadjii-msft commented on GitHub (Dec 6, 2021): This might be related to #5800, or some of the weird line wrapping bugs that I've at least been linking up to that one. I'm thinking particularly the #6901, #10833, #11739 subset.
Author
Owner

@codeofdusk commented on GitHub (Dec 6, 2021):

@zadjii-msft This bug explicitly doesn't repro in wt (conhost only). Please re-label.

@codeofdusk commented on GitHub (Dec 6, 2021): @zadjii-msft This bug explicitly doesn't repro in wt (conhost *only*). Please re-label.
Author
Owner

@zadjii-msft commented on GitHub (Dec 6, 2021):

.... pretty sure I only added the conhost label...
image
image

@zadjii-msft commented on GitHub (Dec 6, 2021): .... pretty sure I only added the conhost label... ![image](https://user-images.githubusercontent.com/18356694/144871416-1938d012-2bb9-49b9-b420-3e62a1e459a8.png) ![image](https://user-images.githubusercontent.com/18356694/144871526-f88dbae4-f28f-482d-b6a2-ec08ec57f67d.png)
Author
Owner

@codeofdusk commented on GitHub (Dec 6, 2021):

Can't see image but it says you added it to the terminal v2.0 milestone. Should be in 22H1 or whichever milestone tracks the new conhost version post-undocking.

@codeofdusk commented on GitHub (Dec 6, 2021): Can't see image but it says you added it to the terminal v2.0 milestone. Should be in 22H1 or whichever milestone tracks the new conhost version post-undocking.
Author
Owner

@zadjii-msft commented on GitHub (Dec 6, 2021):

Ah. I've been kinda re-thinking how we track milestones for our issues recently, since the Windows-specific ones don't totally make sense and have been pretty loosely tracked, and there's not really a difference between the Terminal and Console backlogs anymore. I suppose for the time being, I'll slap that into the 22H1 milestone, but we might be shaking those up in the future, FYI.

Oh hmm we don't even have a 22H1 milestone, but that makes sense, there isn't a 22H1 milestone internally...

@zadjii-msft commented on GitHub (Dec 6, 2021): Ah. I've been kinda re-thinking how we track milestones for our issues recently, since the Windows-specific ones don't totally make sense and have been pretty loosely tracked, and there's not really a difference between the Terminal and Console backlogs anymore. I suppose for the time being, I'll slap that into the 22H1 milestone, but we might be shaking those up in the future, FYI. Oh hmm we don't even have a 22H1 milestone, but that makes sense, there isn't a 22H1 milestone _internally_...
Author
Owner

@codeofdusk commented on GitHub (May 2, 2022):

FYI this repros in the SV2 inbox console. CC @carlos-zamora, @DHowett.

@codeofdusk commented on GitHub (May 2, 2022): FYI this repros in the SV2 inbox console. CC @carlos-zamora, @DHowett.
Author
Owner

@zadjii-msft commented on GitHub (May 3, 2022):

@codeofdusk Yep, looks like this one didn't make it in unfortunately. Hopefully we'll get a chance for the next release.

@zadjii-msft commented on GitHub (May 3, 2022): @codeofdusk Yep, looks like this one didn't make it in unfortunately. Hopefully we'll get a chance for the next release.
Sign in to join this conversation.
1 Participants
Notifications
Due Date
No due date set.
Dependencies

No dependencies set.

Reference: starred/terminal#15950