As a CI artifact in a PR here you get the compiled PHAR so you could easily debug the end result this way.

@ondrejmirtes I must be blind but I don't see the artifact.

in the "compile phar" action

at the bottom

https://github.com/phpstan/phpstan-src/actions/runs/4260429654

sorry for the huge screenshots 😅

@staabm Thanks!

Oke, the generated PHAR now looks like this:

            ksort($fileErrorsCounts, SORT_STRING);
            foreach ($fileErrorsCounts as $message => $count) {
                $php .= '$ignoreErrors[] = [';
                $php .= "\n";
                $php .= "\t'message' => " . var_export(Helpers::escape('#^' . preg_quote($message, '#') . '$#'), \true) . ',';
                $php .= "\n";
                $php .= "\t'count' => " . $count . ',';
                $php .= "\n";
                $php .= "\t'path' => __DIR__ . " . var_export(Helpers::escape($file), \true) . ',';
                $php .= "\n";
                $php .= '];';
                $php .= "\n";
            }

I think this looks good.

I also noticed that the generated baseline did not have a line ending, so I added that.

This is ready.

This pull request has been marked as ready for review.

CS failed: https://github.com/phpstan/phpstan-src/actions/runs/4260736633/jobs/7414242127

When PHP-CS-Fixer is present in the project, what about invoking it and keeping the phpstan PHP file CS always fixed?

Feels like a waste of resources especially on large baselines with 15k lines. PHP CS Fixer is not the fastest....

I will fix the CS in this PR so that we can merge the actual fix.

@ondrejmirtes CS is fixed. And I verified that the new PHAR generates the baseline correctly.

Thank you!