starred/terminal
1
0
Fork 0
mirror of https://github.com/microsoft/terminal.git synced 2026-02-04 05:35:20 +00:00
Code Issues 19.0k Packages Projects Releases 20 Wiki Activity

"Chunked" soft fonts are no longer working correctly #20605

New Issue
Closed
opened 2026-01-31 07:18:52 +00:00 by claunia · 4 comments
claunia commented 2026-01-31 07:18:52 +00:00
Owner
Copy Link

Originally created by @j4james on GitHub (Oct 1, 2023).

Originally assigned to: @lhecker on GitHub.

Windows Terminal version

1.19.2682.0

Windows build number

10.0.19045.3448

Other Software

No response

Steps to reproduce

  1. Download the VT220 soft font from this gist.
  2. Open a cmd shell in Windows Terminal.
  3. TYPE the soft font file downloaded in step 1.

Expected Behavior

The next prompt that is output should be rendered with that low-res VT220 font, looking something like this:

image

Actual Behavior

It appears that some of the characters are only partially downloaded, and some haven't been downloaded at all, so I'm seeing this:

image

If you look at the contents of the font file, you'll see that it's defined as a series of DECDLD sequences, so it's downloading 24 sets of 4 characters, rather than a single 94 character set. I think that's part of the reason for the failure, because fonts that aren't chunked like that don't seem to have the problem.

I also noticed that it only happens in a cmd shell, but not in PowerShell or a WSL bash shell. My guess is that the other shells are buffering their output in a way that hides the problem. It's also possible there is a timing factor, but the cmd shell failure is easily reproducible for me.

And as far as I can tell, the problem was introduced in PR #15991. The commit prior to 41f7ed73c1 was working as expected.

Originally created by @j4james on GitHub (Oct 1, 2023). Originally assigned to: @lhecker on GitHub. ### Windows Terminal version 1.19.2682.0 ### Windows build number 10.0.19045.3448 ### Other Software _No response_ ### Steps to reproduce 1. Download the VT220 soft font from [this gist](https://gist.github.com/j4james/4da454443d8440cadaf84a93f811a35e). 2. Open a cmd shell in Windows Terminal. 3. `TYPE` the soft font file downloaded in step 1. ### Expected Behavior The next prompt that is output should be rendered with that low-res VT220 font, looking something like this: ![image](https://github.com/microsoft/terminal/assets/4181424/5361ea0d-473e-4ffa-8f81-ce1b87c44cb6) ### Actual Behavior It appears that some of the characters are only partially downloaded, and some haven't been downloaded at all, so I'm seeing this: ![image](https://github.com/microsoft/terminal/assets/4181424/b97073d0-0390-4afb-924f-c2e10b30f630) If you look at the contents of the font file, you'll see that it's defined as a series of `DECDLD` sequences, so it's downloading 24 sets of 4 characters, rather than a single 94 character set. I _think_ that's part of the reason for the failure, because fonts that _aren't_ chunked like that don't seem to have the problem. I also noticed that it only happens in a cmd shell, but not in PowerShell or a WSL bash shell. My guess is that the other shells are buffering their output in a way that hides the problem. It's also possible there is a timing factor, but the cmd shell failure is easily reproducible for me. And as far as I can tell, the problem was introduced in PR #15991. The commit prior to 41f7ed73c1e62195919146f689135517aefbaf7b was working as expected.
claunia added the Issue-BugIn-PRArea-VTProduct-Conpty labels 2026-01-31 07:18:52 +00:00
claunia commented 2026-01-31 07:18:54 +00:00
Author
Owner
Copy Link

@j4james commented on GitHub (Oct 1, 2023):

I think I know what the problem is. The new "cork" algorithm causes the VT renderer to flush when the end of the output buffer is reached. This assumes that the VT output is in a ground state, but that's not the case when we're processing a DCS sequence and are in the middle of passing it through to the conpty client. So what's happening is we're flushing a mode ?25 cursor update in the middle of the DCS stream, and that ends the stream early (you can see this in the debug tap).

In theory this could potentially affect all soft fonts, but I think it's more easily triggered by the "chunked" examples I have because they have a line break after each chunk, and the cmd shell's TYPE command probably flushes its output after every line break. Fonts that are written out in a single block will likely only be affected if they're over a certain size. DECCTR palette updates might be affected too, because they also rely on DCS passthrough.

@j4james commented on GitHub (Oct 1, 2023): I think I know what the problem is. The new "cork" algorithm causes the VT renderer to flush when the end of the output buffer is reached. This assumes that the VT output is in a ground state, but that's not the case when we're processing a DCS sequence and are in the middle of passing it through to the conpty client. So what's happening is we're flushing a mode ?25 cursor update in the middle of the DCS stream, and that ends the stream early (you can see this in the debug tap). In theory this could potentially affect all soft fonts, but I think it's more easily triggered by the "chunked" examples I have because they have a line break after each chunk, and the cmd shell's `TYPE` command probably flushes its output after every line break. Fonts that are written out in a single block will likely only be affected if they're over a certain size. `DECCTR` palette updates might be affected too, because they also rely on DCS passthrough.
claunia commented 2026-01-31 07:18:54 +00:00
Author
Owner
Copy Link

@lhecker commented on GitHub (Nov 16, 2023):

I can't figure out how this worked before... VtEngine is still bound into the ~60 "FPS" rendering loop, so why didn't this still randomly happen previously?

@lhecker commented on GitHub (Nov 16, 2023): I can't figure out how this worked before... VtEngine is still bound into the ~60 "FPS" rendering loop, so why didn't this still randomly happen previously?
claunia commented 2026-01-31 07:18:54 +00:00
Author
Owner
Copy Link

@j4james commented on GitHub (Nov 16, 2023):

I assumed there was a lock of some sort that prevented the VtEngine render thread from doing anything while the state machine was busy processing output. If that's not the case then maybe it was a matter of luck that it worked before, and PR #15991 just made the issue more noticeable. But I wasn't able to reproduce the issue prior to that commit, and it happens all the time now.

@j4james commented on GitHub (Nov 16, 2023): I assumed there was a lock of some sort that prevented the `VtEngine` render thread from doing anything while the state machine was busy processing output. If that's not the case then maybe it was a matter of luck that it worked before, and PR #15991 just made the issue more noticeable. But I wasn't able to reproduce the issue prior to that commit, and it happens all the time now.
claunia commented 2026-01-31 07:18:55 +00:00
Author
Owner
Copy Link

@lhecker commented on GitHub (Mar 1, 2024):

This assumes that the VT output is in a ground state, but that's not the case when we're processing a DCS sequence and are in the middle of passing it through to the conpty client.

Edit: I see now (and remember). The problem is not that we flush with the state machine is not in ground state, but rather that VtEngine inserts random stuff into ongoing sequences.

My hope was I can remedy #16769 by doing this, but that doesn't work because it's in ground state during newlines:

void VtIo::CorkRenderer(SCREEN_INFORMATION& screenInfo, bool corked) const noexcept
{
    if (!corked && screenInfo.GetStateMachine().IsGroundState())
    {
        LOG_IF_FAILED(ServiceLocator::LocateGlobals().pRender->PaintFrame());
    }
    _pVtRenderEngine->Cork(corked);
}

Edit 2: It's not that type does line-wise writes, no, it writes lines in chunks of 80 characters. cmd folks were once again clearly ahead of the the game.

@lhecker commented on GitHub (Mar 1, 2024): > This assumes that the VT output is in a ground state, but that's not the case when we're processing a DCS sequence and are in the middle of passing it through to the conpty client. Edit: I see now (and remember). The problem is not that we flush with the state machine is not in ground state, but rather that `VtEngine` inserts random stuff into ongoing sequences. My hope was I can remedy #16769 by doing this, but that doesn't work because it's in ground state during newlines: ```cpp void VtIo::CorkRenderer(SCREEN_INFORMATION& screenInfo, bool corked) const noexcept { if (!corked && screenInfo.GetStateMachine().IsGroundState()) { LOG_IF_FAILED(ServiceLocator::LocateGlobals().pRender->PaintFrame()); } _pVtRenderEngine->Cork(corked); } ``` Edit 2: It's not that `type` does line-wise writes, no, it writes lines in chunks of 80 characters. cmd folks were once again clearly ahead of the the game.
Sign in to join this conversation.
No Branch/Tag Specified
Branches Tags
Labels
Clear labels
⛺ Reserved
A11yCO
A11yMAS
A11ySev1
A11ySev2
A11ySev3
A11yTTValidated
A11yUsable
A11yVoiceAccess
A11yWCAG
Area-Accessibility
Area-AtlasEngine
Area-AzureShell
Area-Build
Area-Build
Area-Chat
Area-CmdPal
Area-CodeHealth
Area-Commandline
Area-CookedRead
Area-DefApp
Area-Extensibility
Area-Fonts
Area-GroupPolicy
Area-i18n
Area-Input
Area-Interaction
Area-Interop
Area-Localization
Area-Output
Area-Performance
Area-Portable
Area-Quality
Area-Remoting
Area-Rendering
Area-Schema
Area-Server
Area-Settings
Area-SettingsUI
Area-ShellExtension
Area-ShellExtension
Area-ShellExtension
Area-Suggestions
Area-Suggestions
Area-TerminalConnection
Area-TerminalControl
Area-Theming
Area-UserInterface
Area-VT
Area-Windowing
Area-WPFControl
AutoMerge
Blocking-Ingestion
Culprit-Centennial
Culprit-WinUI
Disability-All
Disability-Blind
Disability-LowVision
Disability-Mobility
External-Blocked-WinUI3
Fixed
Gathering-Data
good first issue
HCL-E+D
HCL-WindowsTerminal
Help Wanted
Impact-Compatibility
Impact-Compliance
Impact-Correctness
Impact-Visual
In-PR
InclusionBacklog
InclusionBacklog-Windows TerminalWin32
InclusionCommitted-202206
Issue-Bug
Issue-Docs
Issue-Feature
Issue-Feature
Issue-Question
Issue-Samples
Issue-Scenario
Issue-Task
Needs-Attention
Needs-Author-Feedback
Needs-Bisect
Needs-Discussion
Needs-Repro
Needs-Tag-Fix
Needs-Tag-Fix
Needs-Triage
No-Recent-Activity
Priority-0
Priority-1
Priority-2
Priority-3
Product-Cmd.exe
Product-Colortool
Product-Colortool
Product-Colortool
Product-Conhost
Product-Conpty
Product-Meta
Product-Powershell
Product-Terminal
Product-WSL
pull-request
Mirrored from GitHub Pull Request
 Resolution-Answered
Resolution-By-Design
Resolution-Duplicate
Resolution-External
Resolution-Fix-Available
Resolution-Fix-Committed
Resolution-No-Repro
Resolution-Won't-Fix
Severity-Blocking
Severity-Crash
Severity-DataLoss
spam
this-will-be-a-breaking-change
Tracking-External
WindowsTerminal_Win32
Work-Item
zAskModeBug
zInbox-Bug
No Label Area-VT In-PR Issue-Bug Product-Conpty
Milestone
No items
No Milestone
Projects
Clear projects
No project
Assignees
Clear assignees
claunia
No Assignees
1 Participants
Notifications
Due Date
No due date set.
Dependencies

No dependencies set.

Reference: starred/terminal#20605
Delete Branch "%!s()"

Deleting a branch is permanent. Although the deleted branch may continue to exist for a short time before it actually gets removed, it CANNOT be undone in most cases. Continue?