From b35654243af3381266f087cd95e1bfc63da37700 Mon Sep 17 00:00:00 2001 From: Felipe Broering Date: Fri, 17 Apr 2026 18:21:58 -0300 Subject: [PATCH] fix(testkit): make 03_update Log calls alignment-safe MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit The second log line in `Token Count Is Stable After Whitespace Edit` visually aligned `after:` with `before: ` by inserting two spaces between `after:` and the embedded variable. Robot Framework treats two-or-more spaces as an argument separator, so the extra whitespace flipped the statement from Log "after: ${after}[stdout]" # single message into Log "after:" "${after}[stdout]" # message + level and the embedded stdout (a rich count-tokens table) was handed to the `level` slot, which only accepts TRACE/DEBUG/INFO/... That blew up the whole suite for any agent whose corpus is non-trivial enough for `count-tokens` to produce a multi-line report — effectively every production agent. Switch both log lines to `Log Many`, which takes N positional arguments and is immune to how the caller lines them up. Verified locally against ecg-expert (797k-token corpus): suite 03_update now reports 2/2 passed. No other test in the packaged kit has this alignment pattern. Made-with: Cursor --- cli/expert/testkit/suites/03_update.robot | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) diff --git a/cli/expert/testkit/suites/03_update.robot b/cli/expert/testkit/suites/03_update.robot index 174726b..b67b044 100644 --- a/cli/expert/testkit/suites/03_update.robot +++ b/cli/expert/testkit/suites/03_update.robot @@ -30,5 +30,5 @@ Token Count Is Stable After Whitespace Edit ${bumped}= Bump Schema Version ${original} ${copied}= Write Temp Schema ${bumped} ${WORKDIR} ${after}= Run Expert CLI count-tokens --schema ${copied} expect_rc=${None} - Log before: ${before}[stdout] - Log after: ${after}[stdout] + Log Many before: ${before}[stdout] + Log Many after: ${after}[stdout]