Add -V/--version and name the invoked command in the usage output by bertysentry · Pull Request #593 · jawkio/jawk

bertysentry · 2026-08-19T22:15:00Z

Fixes#590, fixes#591.

`-V` / `--version` (#590)

A new standalone mode, recognized like -h only by itself and before the program text:

$ jawk --version
jawk 7.1.00
Java 25.0.3 (Eclipse Adoptium OpenJDK 64-Bit Server VM)

The version comes from the jar manifest's Implementation-Version, which maven-jar-plugin now generates via addDefaultImplementationEntries (the ManifestResourceTransformer carries it into the standalone jar). The Maven pom.properties resource is the fallback, and a plain class directory (IDE runs, unit tests) reports unknown. No build timestamp is involved, so the build stays reproducible.
Combined with any other argument it is rejected, like -h and --list-ext; after the program text it still flows to the script through ARGV, as gawk requires (covered by a test).
Documented in usage(), in the "Help and errors" group and the non-executing modes list of cli-reference.md, and in behavior-changes.md.

Usage names the invoked command (#591)

usage() now prefers the new JAWK_PROGRAM_NAME environment variable over its own jar file name, so the installed launcher's help matches what the user typed:

$ jawk -h
Usage:
jawk [-F fs_val] [-f script-filename] ...
jawk --list-ext

Both launchers pass their own invocation name (${0##*/} in the POSIX launcher, %~n0 in the jawk.cmd shim), on the -jar and the -cp (JAWK_CLASSPATH) invocations alike.
A direct java -jar jawk-x.y.z-standalone.jar -h keeps printing the java -jar ... form, which is correct for that invocation; a blank or unset variable falls back to it too.

Smoke tests

The installer workflow now asserts, on both the POSIX launcher and the Windows shim, that --version reports jawk <version> and that -h names jawk and never mentions java -jar. The steps probe with --version first, so they keep passing while the latest release predates the feature.

Verified end-to-end locally on both launchers (installer run into a scratch prefix, freshly built jar swapped in): --version, -h, plain one-liners, and the JAWK_CLASSPATH code path.

🤖 Generated with Claude Code

Two CLI reporting fixes: - New -V/--version standalone mode: prints "jawk <version>" and the Java runtime, then exits 0. The version comes from the jar manifest's Implementation-Version (now generated via addDefaultImplementationEntries), with the Maven pom.properties resource as fallback and "unknown" when running from a plain class directory. Fixes#590. - usage() now names the command Jawk was invoked as: the launchers written by the one-command installers export their invocation name through the new JAWK_PROGRAM_NAME environment variable, and the CLI falls back to the "java -jar <jar>" form when the variable is not set, which stays correct for direct jar invocations. Fixes#591. The installer smoke tests assert both behaviors through the installed launchers, probing with --version first so they keep passing against releases that predate the feature. Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>

bertysentry · 2026-08-19T23:11:59Z

@codex please review

chatgpt-codex-connector

💡 Codex Review

Here are some automated review suggestions for this pull request.

Reviewed commit:123100a514

ℹ️ About Codex in GitHub

Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you

Open a pull request for review
Mark a draft as ready
Comment "@codex review".

If Codex has suggestions, it will comment; otherwise it will react with 👍.

Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".

chatgpt-codex-connector · 2026-08-19T23:17:03Z

+ AwkTestSupport.TestResult result = AwkTestSupport
+ .cliTest("CLI --version prints the Jawk and Java versions")


Test non-script reporting modes without AwkTestSupport
These new --version tests—and the new help-output tests below—do not run an AWK script, but route through AwkTestSupport.cliTest. The repository explicitly limits that helper to executing AWK scripts and requires ordinary Java behavior to be exercised directly with plain JUnit assertions; instantiate Cli with captured streams and call parse/run directly instead.
AGENTS.md reference: AGENTS.md:L23-L29
Useful? React with 👍 / 👎.

Fixed in 152cae1: the -h/--version reporting-mode tests now drive Cli directly with captured streams and plain JUnit assertions (including the pre-existing -o help test, for consistency). The ARGV pass-through test keeps AwkTestSupport since it runs a script.

chatgpt-codex-connector · 2026-08-19T23:17:03Z

 ## Execution Notes

- `--dump-syntax`, `--dump-intermediate`, `-K`, `-h`, `-?`, and `--list-ext` do not execute the script, and ignore `--persist` and `JAWK_PERSISTENT_MEMORY`. `--profile` does execute it, and keeps the normal AWK output on stdout.
+- `--dump-syntax`, `--dump-intermediate`, `-K`, `-h`, `-?`, `-V`, `--version`, and `--list-ext` do not execute the script, and ignore `--persist` and `JAWK_PERSISTENT_MEMORY`. `--profile` does execute it, and keeps the normal AWK output on stdout.


Distinguish rejected --persist from ignored persistence
For jawk --persist state --version (in either order), parsing rejects the arguments because --version must be used alone, so the command-line --persist option is not ignored as this sentence claims. Only an ambient JAWK_PERSISTENT_MEMORY value is silently irrelevant to a standalone version run; document that distinction so the new CLI behavior is described consistently with line 77 and the implementation.
AGENTS.md reference: AGENTS.md:L35-L39
Useful? React with 👍 / 👎.

Fixed in 152cae1: the sentence now separates the modes that ignore --persist (--dump-syntax, --dump-intermediate, -K) from the standalone modes that reject it as an extra argument, with only an ambient JAWK_PERSISTENT_MEMORY being ignored by all of them.

chatgpt-codex-connector · 2026-08-19T23:17:03Z

+ # command in its help instead of "java -jar ...". Releases up to
+ # 7.1.00 predate --version and JAWK_PROGRAM_NAME, so probe with
+ # --version first and only assert on releases that support it.
+ if out="$("${HOME}/.local/bin/jawk" --version 2>/dev/null)"; then


Stop treating every version failure as a legacy release
In the POSIX installer smoke job here, and equivalently in the Windows shim job, every nonzero --version result takes the legacy-release skip path. Once a release supports this option, a regression that makes --version fail will therefore still leave both smoke jobs green instead of enforcing the feature they were added to test; use the probed release version or another explicit capability boundary to skip only known legacy releases and fail for supported ones.
Useful? React with 👍 / 👎.

Fixed in 152cae1: both jobs now skip only releases <= 7.1.00 (POSIX compares the probed release version with sort -V; Windows reads the installed jar's pom.properties like the pinned-version check) and otherwise require --version to succeed, so a regression fails the job.

- Non-script reporting-mode tests (-h, --version) no longer go through AwkTestSupport, which is reserved for tests that run AWK scripts: they drive Cli directly with captured streams and plain JUnit assertions. - cli-reference.md now distinguishes the modes that ignore --persist (--dump-syntax, --dump-intermediate, -K) from the standalone modes that reject it as an extra argument. - The installer smoke tests skip the version/help assertions only for releases up to 7.1.00 (the last without --version), instead of treating any --version failure as a legacy release, so a regression in a later release fails the jobs. The Windows job reads the installed version from the jar's pom.properties, like the POSIX pinned-version check. Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>

bertysentry · 2026-08-19T23:22:20Z

@codex please review again

chatgpt-codex-connector

💡 Codex Review

Here are some automated review suggestions for this pull request.

Reviewed commit:152cae1b64

ℹ️ About Codex in GitHub

Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you

Open a pull request for review
Mark a draft as ready
Comment "@codex review".

If Codex has suggestions, it will comment; otherwise it will react with 👍.

Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".

chatgpt-codex-connector · 2026-08-19T23:25:37Z

+ # 7.1.00 predate --version and JAWK_PROGRAM_NAME: skip only those,
+ # so a --version regression in a later release fails this job.
+ latest="${{ steps.release.outputs.version }}"
+ if [ "$(printf '%s\n' 7.1.00 "${latest}" | sort -V | tail -n 1)" = "7.1.00" ]; then


Enforce checks for the first supporting release
Checked both new installer smoke-test sections: this cutoff skips v7.1.00, but pom.xml:6 identifies the current development line as 7.1.00-SNAPSHOT, so v7.1.00 is expected to be the first release containing --version and JAWK_PROGRAM_NAME. As a result, regressions in that first supported release take the legacy skip path here and in the Windows [version]...-le check instead of failing the jobs. Fresh evidence beyond the prior thread is the repository's current development version; use the last genuinely legacy release as the cutoff.
Useful? React with 👍 / 👎.

Fixed: the cutoff is now 7.0.01, the last release that actually predates the feature (7.1.00-SNAPSHOT is the current development line, so 7.1.00 will be the first supporting release).

The development line is 7.1.00-SNAPSHOT, so 7.1.00 will be the first release that ships --version and JAWK_PROGRAM_NAME; the last genuinely legacy release is 7.0.01. With the previous cutoff a regression in the first supporting release would have taken the skip path. Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>

bertysentry · 2026-08-19T23:26:45Z

@codex please review again

chatgpt-codex-connector · 2026-08-19T23:29:13Z

Codex Review: Didn't find any major issues. More of your lovely PRs please.

Reviewed commit:843ade0ee9

ℹ️ About Codex in GitHub

Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you

Open a pull request for review
Mark a draft as ready
Comment "@codex review".

If Codex has suggestions, it will comment; otherwise it will react with 👍.

Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".

chatgpt-codex-connectorBot reviewed Aug 19, 2026
View reviewed changes

bertysentry merged commit 0c960b5 into mainAug 20, 2026
8 checks passed

bertysentry deleted the cli-version-and-program-name branch August 20, 2026 11:42

		AwkTestSupport.TestResult result = AwkTestSupport
		.cliTest("CLI --version prints the Jawk and Java versions")

Uh oh!

Conversation

bertysentry commented Aug 19, 2026

-V / --version (#590)

Usage names the invoked command (#591)

Smoke tests

Uh oh!

bertysentry commented Aug 19, 2026

Uh oh!

chatgpt-codex-connectorBot left a comment

Choose a reason for hiding this comment

💡 Codex Review

Uh oh!

chatgpt-codex-connectorBotAug 19, 2026

Choose a reason for hiding this comment

Uh oh!

bertysentryAug 19, 2026

Choose a reason for hiding this comment

Uh oh!

chatgpt-codex-connectorBotAug 19, 2026

Choose a reason for hiding this comment

Uh oh!

bertysentryAug 19, 2026

Choose a reason for hiding this comment

Uh oh!

chatgpt-codex-connectorBotAug 19, 2026

Choose a reason for hiding this comment

Uh oh!

bertysentryAug 19, 2026

Choose a reason for hiding this comment

Uh oh!

bertysentry commented Aug 19, 2026

Uh oh!

chatgpt-codex-connectorBot left a comment

Choose a reason for hiding this comment

💡 Codex Review

Uh oh!

chatgpt-codex-connectorBotAug 19, 2026

Choose a reason for hiding this comment

Uh oh!

bertysentryAug 19, 2026

Choose a reason for hiding this comment

Uh oh!

bertysentry commented Aug 19, 2026

Uh oh!

chatgpt-codex-connectorBot commented Aug 19, 2026

Uh oh!

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

1 participant

`-V` / `--version` (#590)