SA: Update operations.

[drupol]

This PR:

• Fix some typing informations
• Investigated on why Psalm >= 4.9.0 trigger issues with scanLeft and scanRight SA tests. (After bisecting Psalm, the commit introducing the issue is vimeo/psalm@acc7ee2 from PR Fix #6066 - introduce more robust system for capturing template constraints vimeo/psalm#6072). Turns out that it was normal actually. To be confirmed.

[AlexandruGG]

@drupol hey, indeed this looks like an issue with Psalm. Could you reproduce it at https://psalm.dev/ so we can also compare with PHPStan and open an issue?

[drupol]

@drupol hey, indeed this looks like an issue with Psalm. Could you reproduce it at https://psalm.dev/ so we can also compare with PHPStan and open an issue?

No, I haven't had the time to do it yet, I will most probably focus on this in the evening.

[drupol]

@AlexandruGG I think it's a false positive. Actually Psalm was right, the $concat callback has a nullable $carry parameter, I didn't expected that.

Could you please let me know if it's ok for you?

[AlexandruGG]

@AlexandruGG I think it's a false positive. Actually Psalm was right, the $concat callback has a nullable $carry parameter, I didn't expected that.

Could you please let me know if it's ok for you?

Yep you're right!

Comparing to Reduce, the scans will always add the initial parameter passed in, which has a default null. Which means if the Collection is empty then you will be left with a Collection<int, null>. Whereas with Reduce you always get the type returned by the callback, and if the collection is empty it will continue to be empty