Repository navigation
A quoted enabled: "false" in the server-wide config does not turn reporting off #15
Description
A quoted enabled: "false" (or 'false', "off", 'no') in plugins/trace/config.yml does not turn reporting off.
Where: src/main/java/software/stephenson/trace/TraceClient.java. ENABLED_LINE is ^\s*enabled\s*:\s*(\S+). parseServerWideConfig passes the captured group straight to isOff, which trims, lowercases, and compares against off, false, 0 and no. For enabled: "false" the captured group is "false", quotes included, so isOff returns false and the client stays enabled.
The tags: block already goes through scalar(), which strips double and single quotes. The server-id: line does too (server-id: "abc-123" is accepted, see installId_anExistingServerIdIsUsedAndAnUnusableOneIsIgnored). The enabled: line is the only line-reader path that does not.
Why it matters: this is the server-wide opt-out. Operators edit the file by hand, and ci: "true" sits a few lines below in the same file, so quoting the switch is a plausible edit. The result fails open: an operator who believes reporting is off for every plugin still has events sent. Strictly speaking, YAML reads "false" as a string rather than a boolean. But the README and the file's own comment ("Set enabled to false") describe the switch, not YAML typing.
Options for the maintainer (this touches the opt-out contract, so it needs a human decision):
- run the
enabled:value throughscalar()beforeisOff, so quoted spellings disable, the same as for tags andserver-id:; or - keep the current behaviour and document in the README "Opting out" table that the value must be unquoted.
A characterization test pinning the current behaviour (serverWideConfig_aQuotedFalseDoesNotDisableYet) is being added in a Stage B test-expansion PR.
This issue was drafted during a Gardener session (https://github.com/Stephenson-Software/gardener).
drafted by Claude on behalf of Daniel Stephenson
Activity
Cover the line reader's escapes, comments and quoted switch (#16)
dmccoystephenson commentedon Oct 18, 2026
Reopened: this issue was closed automatically, not resolved. The body of PR #16 contained the phrase "should be flipped by whatever PR resolves #15", and GitHub reads "resolves #15" as a closing keyword, so the merge of #16 closed this issue.
The behaviour described here is unchanged on
main(2a2b483).parseServerWideConfigstill passes the rawENABLED_LINEcapture toisOff, and the characterization testserverWideConfig_aQuotedFalseDoesNotDisableYetstill asserts that a quotedenabled: "false"leaves reporting on. The maintainer decision this issue asks for is still open.This comment was drafted during a Gardener session (https://github.com/Stephenson-Software/gardener).
drafted by Claude on behalf of Daniel Stephenson