Store and restore "failed" configuration if init step aborts

[mbg]

This PR mainly contains cherry-picked changes from #4223 for easier review and since they are somewhat separate from the main goal there. In #4223, the goal is to persist and upload diagnostics that are created before the database is initialised. That is the case if, for example, we produce diagnostics and then realise that we should abort the analysis and shouldn't initialise a database.

One issue with making diagnostics from a failed init run available is that the init-post action does not produce or upload any debug artifacts (including the failed SARIF file that would contain the diagnostics) if no Config was written by init. However, init only writes the Config these days if it successfully runs to completion.

This PR addresses this by making the following changes:

• A partial, "failed" Config is written by init if the step is aborted (i.e. an exception is thrown in the first try/catch block). To avoid potentially breaking any code that treats the existence of the normal configuration file as a signal that init was successful, we store the "failed" configuration with a different name.
• Although we only write this "failed" Config when a full Config is available, we write a minimal subset of properties that are needed by init-post to try and produce the debug artifacts we are after. This is also intended to guard against the possibility that some functions interpret the presence of some properties as indication that init was successful.
• The init-post Action reads this "failed" config only if a normal config couldn't be found.
• A new ActionState type is added. This is equivalent to Config by default and, besides having a name which more accurately reflects what Config is, has a type parameter that can be used to mark only some properties as required while unnamed ones are optional.
• That type is then used to better encode the requirements of various Config consumers so that they don't unnecessarily require a full Config.
• This then allows init-post to potentially call them with a "failed" config.

Risk assessment

For internal use only. Please select the risk level of this change:

• Low risk: Changes are fully under feature flags, or have been fully tested and validated in pre-production environments and are highly observable, or are documentation or test only.
• High risk: Changes are not fully under feature flags, have limited visibility and/or cannot be tested outside of production.

Which use cases does this change impact?

Workflow types:

• Advanced setup - Impacts users who have custom CodeQL workflows.
• Managed - Impacts users with dynamic workflows (Default Setup, Code Quality, ...).

Products:

• Code Scanning - The changes impact analyses when analysis-kinds: code-scanning.
• Code Quality - The changes impact analyses when analysis-kinds: code-quality.
• Other first-party - The changes impact other first-party analyses.

Environments:

• Dotcom - Impacts CodeQL workflows on github.com and/or GitHub Enterprise Cloud with Data Residency.
• GHES - Impacts CodeQL workflows on GitHub Enterprise Server.

How did/will you validate this change?

• Unit tests - I am depending on unit test coverage (i.e. tests in .test.ts files).
• End-to-end tests - I am depending on PR checks (i.e. tests in pr-checks).

If something goes wrong after this change is released, what are the mitigation and rollback strategies?

• Feature flags - All new or changed code paths can be fully disabled with corresponding feature flags.
• Rollback - Change can only be disabled by rolling back the release or releasing a new version with a fix.
• Development/testing only - This change cannot cause any failures in production.
• Other - Please provide details.

How will you know if something goes wrong after this change is released?

• Telemetry - I rely on existing telemetry or have made changes to the telemetry.
  • Dashboards - I will watch relevant dashboards for issues after the release. Consider whether this requires this change to be released at a particular time rather than as part of a regular release.
  • Alerts - New or existing monitors will trip if something goes wrong with this change.

Are there any special considerations for merging or releasing this change?

• No special considerations - This change can be merged at any time.

Merge / deployment checklist

• Confirm this change is backwards compatible with existing workflows.
• Consider adding a changelog entry for this change.
• Confirm the readme and docs have been updated if necessary.

[Copilot]

Copilot encountered an error and was unable to review this pull request. You can try again by re-requesting a review.

[Copilot]

Warning

• Copilot's review of this pull request may be incomplete because some of the changed files are excluded by your Copilot content exclusion settings. See Excluding content from Copilot for details.

🟡 Changes recommended

The central aborted-configuration persistence and restoration path lacks direct unit coverage, with additional minor diagnostic and documentation issues.

4 open findings

• Add tests for aborted config persistence and action guards · New
• Correct diagnostic for missing database location · New
• Update documentation link to saveAbortedInitConfig · New
• Use correct Config property name: codeQLCmd · New

Files excluded by content exclusion policy (1)

• lib/entry-points.js

🧠 Review effort: Balanced

[henrymercer]

Some suggestions and comments, but this broadly looks good to me. I'm happy with testing the new functionality here using the PR check added by #4223.

[henrymercer]

OOI, does TypeScript not infer this itself?

[mbg]

If we have e.g.

function bar(config: ActionConfig<"codeQLCmd">) {}

function foo(config: Partial<Config>) {
   if (config.codeQLCmd === undefined) {
      return;
   }

   bar(config); // Type error
}

because TypeScript narrows the type of config.codeQLCmd not config.

[henrymercer]

Should we narrow this log down to "before CodeQL could be set up", now that we have that more restrictive condition?

[mbg]

Firstly, your comment made me realise that this change to the check is no longer needed. I initially allowed undefined for codeQLCmd in AbortedInitConfig, but then realised that this was feature creep for a potential future use where we might want to write the aborted configuration before the CLI is setup. However, since there's currently no need for this, I simplified it to require codeQLCmd to be set but then forgot to revert this check here. I'll do that now.

[henrymercer]

Bikeshed: ConfigWith?

Also, we should be careful with properties that might be optional, e.g. buildMode can be undefined in a successful configuration.

[mbg]

Some time ago we talked about how Config is an awful name for the type and that it's really more representing the state of the Action that we persist across steps. It's also ambiguous as to whether it refers to the Action, CLI, or something else. I thought that I would use this opportunity to move away from that and introduce a better name.