bug: duplicate extmarks from two-pass rendering #143

New issue

Closed

opened 2026-03-04 05:59:47 +00:00 by barrettruth · 0 comments

barrettruth commented

2026-03-04 05:59:47 +00:00

Problem

The decoration provider's two-pass rendering (pass 1: line backgrounds + intra-line; pass 2: deferred treesitter) produces duplicate extmarks on some hunk lines. Observed in a debug dump of email_quote_demo.diff:

Line 30 (> + pass): two DiffsAdd extmarks, two DiffsAddNr, two DiffsAddText
Line 12 (+ pass): same duplication on the non-quoted equivalent

Both the quoted and non-quoted hunks exhibit this, so it's unrelated to email quote parsing — it's a pre-existing issue in the two-pass pipeline.

Pass 2 likely reapplies line backgrounds and intra-line diffs that pass 1 already set, without clearing pass 1's extmarks first (or clearing per-hunk ranges but missing some).

Impact

Functionally harmless (duplicate extmarks with same priority/hl_group are visually identical), but doubles the extmark count unnecessarily and could matter for large buffers with many hunks.

## Problem The decoration provider's two-pass rendering (pass 1: line backgrounds + intra-line; pass 2: deferred treesitter) produces duplicate extmarks on some hunk lines. Observed in a debug dump of `email_quote_demo.diff`: - Line 30 (`> + pass`): two `DiffsAdd` extmarks, two `DiffsAddNr`, two `DiffsAddText` - Line 12 (`+ pass`): same duplication on the non-quoted equivalent Both the quoted and non-quoted hunks exhibit this, so it's unrelated to email quote parsing — it's a pre-existing issue in the two-pass pipeline. Pass 2 likely reapplies line backgrounds and intra-line diffs that pass 1 already set, without clearing pass 1's extmarks first (or clearing per-hunk ranges but missing some). ## Impact Functionally harmless (duplicate extmarks with same priority/hl_group are visually identical), but doubles the extmark count unnecessarily and could matter for large buffers with many hunks.