I have a feeling this commit is causing breakage on PHP 8.5 as the openssl_sign() $algorithm parameter appears to no longer support aliases.

This code:

openssl_sign($signHeader, $signature, $privKey, 'sha256WithRSAEncryption');

Now fails with the following error:

openssl_sign(): Unknown digest algorithm

Found via: https://github.com/PHPMailer/PHPMailer/actions/runs/16728360948/job/47350003979#step:9:33

@jrfnl I highly doubt it. I can run a bisect in half an hour or so though

@nielsdos That would be appreciated!

I agree that based on the code in this PR it seems unlikely, but it was the only commit I could find for PHP 8.5 which related to the openssl_sign() method and the last passing build in PHPMailer is from a couple of days before this was merged and the first failing one after this was merged, so the timing seems right, which is why I left the comment.

@nielsdos Looking more closely - you're right, it can't be this commit - the (failing) CI builds from a PR were confusing the issue.

Last passing is actually on May 31st, first failing on June 18th. Still mystifying though why the alias no longer works. I couldn't find anything in UPGRADING or NEWS related to this.

Shall I open a bug report instead ?

@jrfnl It can also be an environment change, e.g. an update to the OpenSSL library itself. You may open a bug report.

Nope, works in 8.4 breaks in 8.5, I'll bisect...

@nielsdos I've gone through all commits in PHP master between May 31st and June 18th and my current suspicion is that this is the culprit: #18516

I previously already verified via var_export(openssl_get_md_methods(true)); and the alias still exists and the output is not really different between PHP 8.4 and 8.5.

@jrfnl You're correct, bisect finished and points to 2f5ef4d

@nielsdos Thank you for doing that and confirming! In the mean time, I've set up a phpt test for alias support in openssl_sign() - I'm waiting to verify the test works as intended and if so, I'll include it in the bug report.

@nielsdos I've opened the ticket now: #19369

Great job @nielsdos for catching this memory leak. Are you planning to opensource the experimental static analysis tool that you used to catch this one?

Great job @nielsdos for catching this memory leak. Are you planning to opensource the experimental static analysis tool that you used to catch this one?

That's the plan eventually, the paper describing the analysis tool is currently under review

Great job @nielsdos for catching this memory leak. Are you planning to opensource the experimental static analysis tool that you used to catch this one?

That's the plan eventually, the paper describing the analysis tool is currently under review

Okay, where can I get my hands/eyes on the pre-release? Is it something online I can find?

Okay, where can I get my hands/eyes on the pre-release? Is it something online I can find?

It's not public yet, it will only become public once the journal publishes the article and source code (after it got reviewer approval).

Okay @nielsdos, thanks for the update. Looking forward.