Avoid using unsafe sprintf()

[alexandre-daubois]

Internals book explicitly specifies that sprintf() should be avoided where possible for safer alternatives.

[nielsdos]

I did a round on this once a long time ago, strange I missed these, thanks for catching this

[TimWolla]

The FPM code is not easy to understand, but I believe it is correct.

[iluuu1994]

Looks correct to me. The fpm code is definitely a readability improvement. Waiting to see if Jakub wants to review, otherwise we can merge in a day or two. Thanks!

[TimWolla]

I thought a little more about the FPM code. Suggestions are not tested, but I've tried to carefully explain my train of thought.

[alexandre-daubois]

Your suggestions are correct, thank you Tim!

[TimWolla]

Double-checked the FPM logic again and it LGTM.

[TimWolla]

@php/release-managers-85 This probably is halfway between bugfix and feature, so please take a look. You can “Squash and merge” directly if you are good with it, no NEWS necessary and the author can't merge themselves.

[bukka]

I think it would be best to separate the FPM changes. I'm really not sure that part should go to 8.5 as it needs proper testing once corrected.

[bukka]

This is master process code - it should not use zend alloc. Ideally all master code should also not immediately exit during the alloc failure so it should keep using standard malloc / realloc / free. It means even the persistent version would not be preferred here. I would prefer not the use smart string at all here