A new pane with no room under the last goes last all the same

A pane from pane/new or New takes the bottom half of its column's last
pane, as acme's coladd does (cols.c:62-65); when that pane is too short
to give it a tag and 2 rows, the fallback halved the tallest pane and
put the new one under it, mid-column (the 7 9 8 found). It now goes last
all the same, with half the tallest pane's rows (its tag measured at the
column's width). Tty from a pane's ctl already follows acme's
makenewwindow (the active column first, util.c:456-467); fs.md now says
so rather than beside that pane.

Co-Authored-By: Claude Opus 5.5 <[email protected]>

1 files changed, 10 insertions, 0 deletions

diff --git a/src/ninep/tree.zig b/src/ninep/tree.zig
index 27091112..af1c589d 100644
--- a/src/ninep/tree.zig
+++ b/src/ninep/tree.zig
@@ -1560,3 +1560,13 @@ test "an a, c or i text cut short through exec is an err in the log, not a msg"
     try testing.expect(!th.logHas(p, "msg 1 Edit: a, c or i"));
     try testing.expectEqualStrings("abc\n", p.panes[0].?.file.?.content);
 }
+
+test "a pane made from pane/new goes last in its column, even when the last pane has no room to halve" {
+    const p = try th.withFile(testing.allocator, "x\n");
+    defer p.deinit();
+    for (0..3) |_| {
+        const made = call(p, .{ .tag = 1, .op = .open, .node = @intFromEnum(TopFile.new) }).reply.handle;
+        p.sync();
+        try testing.expectEqual(made, p.panes[p.col_panes[0][p.col_n[0] - 1]].?.serial);
+    }
+}