fix(desktop): OpenCode Go mimo vision + Windows DPI repaint / 会话切模型读图与 Windows 缩放修复 - #7822
Open
SivanCola wants to merge 1 commit into
Open
fix(desktop): OpenCode Go mimo vision + Windows DPI repaint / 会话切模型读图与 Windows 缩放修复#7822SivanCola wants to merge 1 commit into
SivanCola wants to merge 1 commit into
Conversation
Problem 1. Mid-session switches from a text-only default (e.g. opencode-go/ deepseek-v4-flash) to opencode-go/mimo-v2.5 still treated images as path-only text (esengine#7798). New sessions with mimo as default often worked because they resolved a different vision-capable provider. 2. Windows minimise/restore and restart can leave a stale WebView2 scale layer, showing blown-up or blurry UI (esengine#7794, esengine#7799). Root cause - OpenCode Go stock VisionModels listed only kimi-k3, so EffectiveVision stayed false for mimo-v2.5 and inputImages skipped payloads. - Existing installs kept the old vision list; no migration added mimo. - WebView2 compositor layers were not forced to re-rasterize on restore. Fix - Add mimo-v2.5 to opencode-go vision catalog; migrate stock kimi-k3-only lists on load while preserving explicit empty/custom vision lists. - On Windows visibility restore, force a one-frame compositor repaint. - Regression tests for catalog, migration, mid-session image attach, and the DPI repaint helper. Verification - go test ./internal/config/ ./internal/control/ - tsx desktop/frontend/src/__tests__/dpi-repaint.test.tsx
github-actions
Bot
added
v2
Open
6 tasks
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Summary
Addresses two high-signal items from the 48h bug triage:
mimo-v2.5did not attach vision image payloads when the model is selected under OpenCode Go.Vision (#7798)
opencode-golistedmimo-v2.5as a chat model but not inVisionModels(onlykimi-k3).opencode-go/mimo-v2.5→imageInputEnabledfalse → path text only.mimo-v2.5default often hit the official MiMo provider (vision OK), which made the gap look session-scoped.Fix: add
mimo-v2.5to the OpenCode Go vision catalog; migrate stockvision_models = ["kimi-k3"]lists on load (preserve explicit empty/custom lists).Windows DPI (#7794 / #7799)
visibilitychange → visiblefor Windows to clear stale scale layers after minimise/restore.Test plan
go test ./internal/config/ ./internal/control/tsx desktop/frontend/src/__tests__/dpi-repaint.test.tsxopencode-go/deepseek-v4-flash, switch toopencode-go/mimo-v2.5, paste image → model describes imageRelated
中文摘要
修复 OpenCode Go 下会话内切到 mimo-v2.5 读图失效，并在 Windows 最小化恢复时强制重绘以缓解缩放/模糊。