Conversation
A caller asking for a key that is not there was told "Object by hash code 96354 from given key does not exists": broken grammar, and a number they never saw in place of the key they hold. The message names the key now, as text where its bytes decode as UTF-8 and in hexadecimal where they do not, and a test pins it. Co-Authored-By: Claude Opus 5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01NxFBB1HWHS1xgdM6bT9Qvw
|
These counts changed in this branch,
They are defined in .github/workflows/counts.yml |
|
Two things a reviewer will want to know before reading the diff. The red The step runs with PR #8012 covers the same issue and was opened ahead of this one. Worth picking one and closing the other rather than reviewing both. Generated by Claude Code |
|
The Generated by Claude Code |
|
qulice is red on this PR. Not merging until CI is green. Generated by Claude Code |
|
Not merging automatically: the Generated by Claude Code |
|
Holding off on merging — the Generated by Claude Code |
|



Closes #7988.
A caller who asked a map for a key that is not there was told
— broken grammar, and a polynomial hash the caller never saw standing in for the key they hold. The message is now
The key three is not in the map, built fromkbts, the bytes the map already dataized the key into.Those bytes are printed with
%swhere they decode as UTF-8 and with%xwhere they do not, the two arms wrapped inrecovered:%sterminates on bytes that are not UTF-8, and that termination is what selects the hexadecimal form. So a key that is not text still reports something the caller can recognise.can-format-the-error-message-of-an-absent-keypins the text, reading the reason back through thecant-getfallback withI, the waystring.at's message tests do. It asserts the new wording, so it cannot pass againstmaster.TestEOmapis green (22 tests) on a localmvn clean test -pl :eo-runtime -Deo.deadline=3600.Generated by Claude Code