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

Draw the background separately from the foreground during render passes #8585

New Issue
Closed
opened 2026-01-31 01:33:09 +00:00 by claunia · 9 comments
claunia commented 2026-01-31 01:33:09 +00:00
Owner
Copy Link

Originally created by @DHowett on GitHub (May 26, 2020).

preliminary work: https://github.com/microsoft/terminal/tree/dev/duhowett/background-atlas

I've been thinking a bit about the DX renderer drawing a "background atlas" before it draws any text.

If we separate background drawing into a totally separate phase, we get the following benefits:

  • We do not need to break runs of text until their foreground presentation changes (performance)
  • We will not break ligatures when the background changes (correctness) (fixes #800)
  • We can draw stuff behind the text (visuals, correctness) (touches mini-spec #6151)

It would look, roughly, like this:

step 1 (empty renderer)

image

step 2 (drawing background atlas)

image

step 3 (drawing foreground)

image

may need a mini-spec

Originally created by @DHowett on GitHub (May 26, 2020). preliminary work: https://github.com/microsoft/terminal/tree/dev/duhowett/background-atlas I've been thinking a bit about the DX renderer drawing a "background atlas" _before it draws any text_. If we separate background drawing into a totally separate phase, we get the following benefits: * We do not need to break runs of text until their _foreground presentation_ changes (performance) * We will not break ligatures when the background changes (correctness) (fixes #800) * We can draw stuff behind the text (visuals, correctness) (touches mini-spec #6151) It would look, roughly, like this: ### step 1 (empty renderer) ![image](https://user-images.githubusercontent.com/189190/82858125-d0facf80-9ec7-11ea-9ad5-c89028d5a3c7.png) ### step 2 (drawing background atlas) ![image](https://user-images.githubusercontent.com/189190/82858147-db1cce00-9ec7-11ea-91fc-07f18ee4fa2c.png) ### step 3 (drawing foreground) ![image](https://user-images.githubusercontent.com/189190/82858106-c6d8d100-9ec7-11ea-8eb8-260b10bbf7e6.png) _may need a mini-spec_
claunia added the Area-RenderingIssue-TaskNeeds-Tag-FixProduct-Terminal labels 2026-01-31 01:33:10 +00:00
claunia commented 2026-01-31 01:33:11 +00:00
Author
Owner
Copy Link

@DHowett commented on GitHub (May 26, 2020):

Hey, it works!

image

@DHowett commented on GitHub (May 26, 2020): Hey, it works! ![image](https://user-images.githubusercontent.com/189190/82864397-5044cf00-9ed9-11ea-9f4b-40c3f2e36268.png)
claunia commented 2026-01-31 01:33:11 +00:00
Author
Owner
Copy Link

@DHowett commented on GitHub (May 26, 2020):

Right now, I'm inefficiently iterating over the entire dirty region twice -- once row by row x attr by attr with an AttrRowIterator, and once row by row x cell by cell with a TextBufferCellIterator.

They're the same expense, roughly, because AttrRowIterator doesn't have any affordances for getting the run length and skipping it. If it did, we could dispense with some of the complexity.

IDEALLY, I would have one function that turned out a vector of clusters and a vector of backgrounds. This would save the cost of iterating twice.

@DHowett commented on GitHub (May 26, 2020): Right now, I'm inefficiently iterating over the entire dirty region twice -- once row by row x attr by attr with an AttrRowIterator, and once row by row x cell by cell with a TextBufferCellIterator. They're the same expense, roughly, because AttrRowIterator doesn't have any affordances for getting the run length and skipping it. If it did, we could dispense with some of the complexity. _IDEALLY_, I would have one function that turned out a vector of clusters and a vector of backgrounds. This would save the cost of iterating twice.
claunia commented 2026-01-31 01:33:11 +00:00
Author
Owner
Copy Link

@zadjii-msft commented on GitHub (May 26, 2020):

Okay, so this might be working better than my hack to manually break the run with the cursor on it. I need to add at least O(Rows+Cols) checks to painting the frame in my prototype, to check if we're on the cursor row, and the perf so far is horrifyingly bad.

If this truly doesn't double the work done for a single frame, then this seems like it might be a better solution to combine with the pre/post rendering of the cursor.

@zadjii-msft commented on GitHub (May 26, 2020): Okay, so this might be working better than my hack to manually break the run with the cursor on it. I need to add at least O(Rows+Cols) checks to painting the frame in my prototype, to check if we're on the cursor row, and the perf so far is horrifyingly bad. If this truly doesn't _double_ the work done for a single frame, then this seems like it might be a better solution to combine with the pre/post rendering of the cursor.
claunia commented 2026-01-31 01:33:11 +00:00
Author
Owner
Copy Link

@egmontkob commented on GitHub (May 26, 2020):

VTE does 2-pass rendering like this.

IMO the biggest advantage is that letters can slightly overflow to neighboring cells, e.g. diacritics above or below don't get cropped, bold letters don't get their right edge cut off.

@egmontkob commented on GitHub (May 26, 2020): VTE does 2-pass rendering like this. IMO the biggest advantage is that letters can slightly overflow to neighboring cells, e.g. diacritics above or below don't get cropped, bold letters don't get their right edge cut off.
claunia commented 2026-01-31 01:33:11 +00:00
Author
Owner
Copy Link

@DHowett commented on GitHub (May 27, 2020):

Alright, I've got this down to a single function that takes each dirty region and produces...

  • A list of damaged backgrounds to repaint
  • A list of colored cluster lists to repaint

I don't like the number of vectors here, but... we're saving money by invalidating narrowly whenever possible.

Dirty Region [1..n]
- Origin (SCREEN CHAR COORDS)
- At line wrap point
- Region intersected right half of DBCS char (left trim)
- Background Run [1..n]
  - Attribute
  - Rect Covered (SCREEN CHAR COORDS)
- Colored Foreground Run [1..n]
  - Attribute
  - Columns
  - Cluster [1..n]

Right now, because of some minor issues (#2661), backgrounds are attributes and actual final colors. It's not pleasant!

To fix:

  • reverse is occasionally (!) broken
  • render engines need flags to say "plz no background splitting" (vt engine needs to update drawing brushes per text run, foreground and background)
  • need to re-enable space optimization (if the cell is non-printing (and has no underline, etc.), no point in forwarding it to the cluster renderer!)
@DHowett commented on GitHub (May 27, 2020): Alright, I've got this down to a single function that takes each dirty region and produces... * A list of damaged backgrounds to repaint * A list of colored cluster lists to repaint I don't like the number of vectors here, but... we're saving money by invalidating narrowly whenever possible. ``` Dirty Region [1..n] - Origin (SCREEN CHAR COORDS) - At line wrap point - Region intersected right half of DBCS char (left trim) - Background Run [1..n] - Attribute - Rect Covered (SCREEN CHAR COORDS) - Colored Foreground Run [1..n] - Attribute - Columns - Cluster [1..n] ``` Right now, because of some minor issues (#2661), backgrounds are attributes _and_ actual final colors. It's not pleasant! To fix: * [ ] reverse is occasionally (!) broken * [x] render engines need flags to say "plz no background splitting" (vt engine needs to update drawing brushes per text run, foreground _and_ background) * [ ] need to re-enable space optimization (if the cell is non-printing (and has no underline, etc.), no point in forwarding it to the cluster renderer!)
claunia commented 2026-01-31 01:33:12 +00:00
Author
Owner
Copy Link

@DHowett commented on GitHub (May 27, 2020):

branch

@DHowett commented on GitHub (May 27, 2020): [branch](https://github.com/microsoft/terminal/tree/dev/duhowett/background-atlas)
claunia commented 2026-01-31 01:33:12 +00:00
Author
Owner
Copy Link

@DHowett commented on GitHub (Jun 3, 2020):

Michael's got some ideas that will be less invasive than this.

@DHowett commented on GitHub (Jun 3, 2020): Michael's got some ideas that will be less invasive than this.
claunia commented 2026-01-31 01:33:12 +00:00
Author
Owner
Copy Link

@DHowett commented on GitHub (Jul 1, 2020):

preliminary work: https://github.com/microsoft/terminal/tree/dev/duhowett/background-atlas

@DHowett commented on GitHub (Jul 1, 2020): preliminary work: https://github.com/microsoft/terminal/tree/dev/duhowett/background-atlas
claunia commented 2026-01-31 01:33:12 +00:00
Author
Owner
Copy Link

@DHowett commented on GitHub (Jun 29, 2022):

Atlas supersedes this.

@DHowett commented on GitHub (Jun 29, 2022): Atlas supersedes this.
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-Rendering Issue-Task Needs-Tag-Fix Product-Terminal
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#8585
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?