Sergei Morozov

> I think there is more to that as if I recall correctly the data of creation is might be added somewhere as a header or metadata […] Definitely. The...

As a temporary workaround, one could use the following command in the project directory to patch the package with the proper versions file: ```sh $ php -dphar.readonly=0 -- /path/to/package.phar

I'm not using [composer/package-versions-deprecated](https://packagist.org/packages/composer/package-versions-deprecated) directly. I use [jean85/pretty-package-versions](https://packagist.org/packages/jean85/pretty-package-versions) which switched from the original library ([ocramius/package-versions](https://packagist.org/packages/ocramius/package-versions)) to the fork in `1.3` (see https://github.com/Jean85/pretty-package-versions/pull/13). Unless I did something wrong, when I tried...

> Wouldn't a better solution to migrate towards the Ocramius one? As far as I can tell, Composer 2 is [taking over](https://github.com/composer/composer/blob/d89342dc434d52c88e0e06ce3982da739a467f13/src/Composer/InstalledVersions.php) the installed versions management.

@villfa, thank you for the research and the suggestion. The version detection indeed works this way but the resulting package now contains the`vendor` directory files that belong to the development...

> It's a bit annoying actually since it's not files you want to include in the PHAR The same is about the development dependencies but somehow Box figures out that...

> The alternative obviously is to have the version dumper working without git... But that sounds like a lot more work That sounds unrealistic. The version of the root package...

Another approach might be to provide some backchannel from the mutated code to Infection and signal about the non-effective branch. E.g. call some Infection API from the mutated code (if...

I don't think so. The approach in the description allows white-listing a known class of problems w/o polluting code. Furthermore, with the comments approach, maintenance of this code will require...

> PHPStan can't cater to all the specific situations in the world. Fair point. > It shouldn't make the messages less readable just so they could be ignored. I wouldn't...