-
Notifications
You must be signed in to change notification settings - Fork 8.3k
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Accessibility regressions in the 21H1 inbox console #9239
Comments
Cc @carlos-zamora. |
Can you check the file version (right click, properties) for conhost.exe? I am not sure how NVDA’s Certainly we have made 0 changes to conhost in 21H1. I wasn’t even aware there was a 21H1 publicly available. |
Version number is 10.0.19641.746 |
That doesn’t seem right. I’ll need to check the build when I get back to the office tomorrow, but 19641 corresponds to a version of Windows from two development cycles ago, numerically, and would not contain any of our recent UIA fixes. Just so you know: 21H1 is a patch release from the servicing team, and they did not contact us about releasing any conhost changes. I have no reason to believe that the code in conhost 21H1 is any different from 20H1, which was the last non-patch upgrade that brought in a whole new tree of Windows code indiscriminately. |
The versions of Windows that take your computer down for 10-75 minutes and reboot three times tend to contain the entire development tree from the OS repository before they’re snapped. The builds that increment the UBR or patch version, or the ones that increment the main version by 1 (so, from 19041 to 19042 or 19043) are servicing updates that have hand-selected features backported into them. Since we were not hand-selected for those releases, no conhost changes are present in 20H2 or 21H1. |
Discussed this with @codeofdusk off GitHub. The changes we were targeting for "21H1" did not get selected for the build that was released in the early half of 2021. We'll regroup here, but there's no regression per se. Just a lack of ... progression? I guess? |
Fix for microsoft/terminal#9239 Summary of the issue: Windows 10 21H1 has been released to insiders with significantly reduced scope (i.e. excluding the UIA changes it was originally set to include), making our internal names misleading. Additionally, there are plans to "undock" conhost from Windows, removing the guarantee that a given Windows version will run a particular version of the console (it's technically already possible to run newer console on older Windows, I do it for testing). Description of how this pull request fixes the issue: Renames: - NVDAObjects.UIA.winConsoleUIA.is21H1Plus -> NVDAObjects.UIA.winConsoleUIA.isImprovedTextRangeAvailable - NVDAObjects.UIA.winConsoleUIA.consoleUIATextInfo -> NVDAObjects.UIA.winConsoleUIA.ConsoleUIATextInfo (Start class name with upper case) - NVDAObjects.UIA.winConsoleUIA.consoleUIATextInfoPre21H1 -> NVDAObjects.UIA.winConsoleUIA.ConsoleUIATextInfoWorkaroundEndInclusive - The implementation works around both end points being inclusive (in text ranges) before microsoft/terminal#4018 - Workarounds for expand, collapse, compareEndPoints, setEndPoint, etc Also moves the text override to the workaround textInfo as #10036 is no longer reproducible in new console. Co-authored-by: Reef Turner <reef@nvaccess.org>
Environment
Steps to reproduce
nav.is21H1Plus
Expected behavior
True
is returned, as in OpenConsole commit 12b12d5.Actual behavior
False
is returned, as in 20H2 inbox, indicating a regression from #4495.Additional context
For testing, I've replaced conhost with OpenConsole on my system, but reverted to inbox before applying the 21H1 update. Perhaps Windows Update didn't install the new inbox conhost on my system?
The text was updated successfully, but these errors were encountered: