Feature request: CLI option to clear the results cache

[jrfnl]

Repost from squizlabs/PHP_CodeSniffer#2993:

While testing #2992, I really missed an option to clear the cache, but allow for it to be set.

The --no-cache option only tells PHPCS to not use the cache for that particular run.

It doesn't clear an existing cache file, or in case of the temp directory being used: clear all existing cache files, as those can accumulate to be quite a large number of files quickly.

Please consider this a feature request for a CLI argument to clear the cache(s) to be added.

---

[gfaugere]

Hey, does this MR implementing a --delete-cache option (squizlabs/PHP_CodeSniffer#3134) still holds any relevance?

If not I'll close it to reduce bloat

[jrfnl]

@gfaugere I still have to find the time to go through all remaining open PRs in the old repo to see which ones are viable and invite the creators to recreate the PR here. Sorry I haven't got round to that yet.

If you think that PR addresses this issue, feel free to recreate it in this repo. It may still need changes after review, but I will definitely take a look at it.

[jrfnl]

Some notes about solving this (@gfaugere which you may want to have a look at):

• The cache option will either cache to a named file or to a file in the system temp directory.
• We may need to have a good think about what should be cleared when / how the clear cache option should work:
  • If a predefined cache name is included in the command to clear the caches, clear only that.
  • If the phpcs.xml[.dist] file contains a predefined cache name to always (re)use, clear only that ?
  • If neither of the above applies, how will it be determined which cache file to clear ? Search the project directory for a file which looks like a PHPCS cache file ? Will all caches in the system temp directory be cleared ? If the latter, this will not only affect the project for which PHPCS is currently being run, but will also clear the PHPCS caches for all other projects using PHPCS on the end-users computer. This may not be the intention.

All in all, we may need to have a discussion about the design of the feature before looking at an implementation.