fix redrawing & vanishing last column - #145
Merged
Merged
Conversation
writer.writeAll(ansi.cursor_home) prints the byte \x1b[H (go to row 1, column 1 but ommit params) I changed it to: ansi.cursorTo0(writer, @intcast(line_count), 0) which prints the byte \x1b[1;1H (go to row 1, column 1, exclicit params) (it actually prints \x1b[{row+1};1H, but you get the idea)
Contributor
Author
|
This is my first Zig PR ever. |
Addressing rows absolutely removes the scroll that meszmate#143 reported, and skipping `EL` on a line that fills the row removes the erased border that meszmate#144 reported. Both need bounds the original change did not apply. A `CUP` row is a `u16`, and the terminal clamps it to the screen: * A view of 65536 lines or more overflowed the row counter and panicked. `writeFull` is the path that absorbs oversized frames, so the crash sat on exactly the input it exists to handle, and `.full` mode never bounds the line count at all. * Rows past the bottom clamped onto the last row, so each overflow line overwrote the one before it: six lines on a three-row screen left `a`, `b`, `f` rather than the top three. Count those lines, so `last_line_count` still describes the frame, but do not draw them. `writeDiff` bounds its trailing clear the same way. It cannot reach a `last_line_count` past the screen today — an oversized frame marks the renderer dirty, which routes the next frame through `writeFull` — but two paths clearing to different limits is a trap, not an invariant. The erase predicate needs two corrections of its own, both in `fillsRow`: * `width < size.width` is never true when the width is zero, which suppressed every erase and left stale text on screen. `TIOCGWINSZ` reports 0x0 for a terminal whose window size was never set, and the 80x24 fallback in `getSize` only covers the non-TTY case. An unknown width says nothing about which lines fill a row, so erase. * A line that wraps returns to the edge at every multiple of the row width, not only when it fills one row exactly. Ten columns of text on a five-column screen parks the cursor on the edge a second time, where `EL` eats the last character just the same. Test the width modulo the row rather than against it.
Two gaps let the renderer's own regressions through. `VirtualScreen` counted bytes, not columns, and knew nothing of the screen it was modelling. It could not represent the bug behind meszmate#144 at all: a character written to the last column parks the cursor there with its wrap deferred, and `EL` erases from the column the cursor occupies — taking that character with it. A frame drawn to the full width of the screen modelled perfectly and still lost its right-hand border on a real terminal. Model columns, autowrap, wrap-pending and `CUP` clamping, so the harness can show what a terminal would. `\x1b[H` used to be proof of a full repaint, because only `writeFull` emitted it. Absolute addressing made it `\x1b[1;1H`, which `writeDiff` also emits whenever row 0 is the row that changed. Every rewritten assertion still discriminated by luck — each left row 0 untouched — but they no longer tested what their names claimed. Assert instead that a row which did *not* change was rewritten, which is the actual difference between the two paths. Covers what neither path had: a frame that fills the width, a wrapped line that ends on the edge, truncation past the bottom row, a view too tall for a `u16` row, and an unknown terminal width.
meszmate
self-requested a review
August 28, 2026 11:44
Owner
|
Thanks, and for the two issues too. Both diagnoses were spot on, which is the hard part. Pushed two commits on top:
589/589 pass, green in all optimize modes. Closes #143, closes #144. |
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
2 participants
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.
closes #144 #143