Fix pcntl get/set priority

[juan-morales]

Code

Extension: pcntl
Functions: pcntl_setpriority() , pcntl_getpriority()

The problem

The PHP code to handle the errors for the mentioned functions is Linux aware only. Mac OS has a different behavior and uses the same error code for different things; plus FreeBSD does not respond in the same way as Linux or Mac under certain circumstances.

Because this feature (managing processes) is very OS dependent we cannot fully test the whole code, and also we cannot manage it in a uniform way (one code for all of them).

Documentation per OS related to possible error codes that can be return for the getpriority() and setpriority() functions

Errors per OS:
Mac

FreeBSD

Linux

The PHP code at the moment (master)

pcntl_getpriority()

pcntl_setpriority()

So as we can see the code to manage the errors is wrong as it will work as expected in Linux and FreeBSD, but not in MacOS.

This change pretends to fix this issue.

Relates to (already closed):
#8994
#8993

This is the first time I do such a change, please keep that in mind and advise me about anything that should be change, delete, adjust, etc.

Thanks in advance to all reviewers.

@cmb69 as promised ... here it is.

[juan-morales]

(is really hard to test it before pushing when you don't have a Mac)

[cmb69]

Thank you for the PR! I'll try to do a full review tomorrow, but left some comments for things that caught my attention while skimming over the patch.

[cmb69]

Preprocessor directives should always start at the beginning of a line (there may be whitespace between the # and the symbol, but in this case this would not be helpful:

Suggested change

	#ifdef PRIO_DARWIN_BG
	#ifdef PRIO_DARWIN_BG

Same for the other new directives in this file.

[juan-morales]

Thank you for the PR! I'll try to do a full review tomorrow, but left some comments for things that caught my attention while skimming over the patch.

Thanks @cmb69 a lot! I will wait for your full review before I do the suggested changes (so the pipeline does not have to run again)

[bukka]

The logic makes sense to me. Just those small NITs from @cmb69 .

[juan-morales]

The logic makes sense to me. Just those small NITs from @cmb69 .

Thanks for the review; and yes, I will apply the NITs from @cmb69 after the full review.

PD: I dont know what is better for you people, if

1. Keep the code as it is so I can get a feedback as rich as possible, and work it out from there
   or
2. I fix the NITs now and review it again later

I do whatever is better for the process

[bukka]

I just did the full review ;)

[bukka]

There's not that much to review though excepting checking the docs and logic.

[bukka]

Please fix those NITs, squash it to a single commit and push force it here.

[juan-morales]

@cmb69 @bukka change ready. But TRAVIS is failing ,according to the logs I saw is failing in a test that is not from my change. Can you advice on this?

[cmb69]

That Travis CI failure is intermittent, so not related to this PR.

Thank you for the patch!