signsofai --version has said 0.1.0 since 0.1.0 - #84
Merged
Merged
Conversation
`Program.cs` restated the version as a constant and nobody ever bumped it. The package has gone out as 0.2.0 through 0.7.0 while the tool told anyone who asked that they were running the first release ever published. That is the failure mode from 24 August, one channel over. A support message then said "desktop 0.4.0 is not published" because the desktop app displayed no version at all and the reporter could not tell which build they had. #64 fixed that one and added a guard. Nobody looked at the CLI. It is worse here than there. The desktop said nothing; the CLI says something false. Someone checking whether they already have today's em-dash fix runs `--version`, reads 0.1.0, and concludes they are seven releases behind when they are current. `BuildInfo` reads the assembly the compiler stamped from `<Version>`, so the two cannot disagree — the same reasoning `DesktopRelease.ZipUrl` uses to interpolate its URL instead of restating it. The guard runs the tool that was just built and compares what it says against what the csproj packs. Verified by mutation: with the constant put back, the step fails with 0.1.0 against 0.7.0. A test asserting the source does not contain a literal would have been satisfied by a comment. Found while verifying that 0.7.0 on NuGet actually carried the em-dash fix. It does — the learner's essay comes back with no em-dash finding from the published package. The version it reported while proving it was 0.1.0. Co-Authored-By: Claude Opus 5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_015PEbbiYSNPw7jE3LrPNhyF
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
1 participant
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.
src/SignsOfAI.Cli/Program.csrestated the version as a constant and nobody ever bumped it:The package has shipped as 0.2.0 through 0.7.0 while the tool told anyone who asked that they were running the first release ever published.
This is the 24 August failure, one channel over
A support message then said "desktop 0.4.0 is not published" — because the desktop app displayed no version anywhere and the reporter could not tell which build they had. #64 fixed that one and added a guard. Nobody looked at the CLI.
It is worse here than there. The desktop said nothing; the CLI says something false. Someone checking whether they already have today's em-dash fix runs
--version, reads0.1.0, and concludes they are seven releases behind when they are current.The fix
BuildInforeads the assembly the compiler stamped from<Version>, so the two cannot disagree — the same reasoningDesktopRelease.ZipUrlalready uses to interpolate its URL rather than restate it.The guard
A CI step runs the tool that was just built and compares what it says against what the csproj packs. Verified by mutation: with the constant put back, the step fails with
0.1.0against0.7.0.It deliberately is not a test asserting the source contains no literal — that class of test is satisfied by a comment, which is the lesson from the boot tests on 31 August.
How it was found
While verifying that 0.7.0 on NuGet actually carried the em-dash fix. It does — installed from NuGet, the learner's essay comes back with no em-dash finding. The version it reported while proving that was
0.1.0.Release
Coreis untouched, so no score changes and the calibration is not affected. This isClionly. Cutting 0.7.1 is a judgement call: the fix matters most to somebody diagnosing whether they have 0.7.0, and right now the tool answers that question wrongly.🤖 Generated with Claude Code
https://claude.ai/code/session_015PEbbiYSNPw7jE3LrPNhyF