php 5.6 compatibility

[Maks3w]

No description provided.

[DASPRiD]

What exactly can go wrong here in 5.6?

[Maks3w]

"[math][php-56] PHP now throws E_WARNING when string is not a valid gm… …
…p integer."

[Ocramius]

@DASPRiD see https://travis-ci.org/zendframework/zf2/jobs/20643164#L695

@Maks3w is error suppression actually how we're supposed to deal with this?

[Maks3w]

I thought in throw InvalidArgumentException but interface contract explictly use false for errors handling.
$res still having false as value.

[DASPRiD]

For that error handler setting, don't we have something in Stdlib?

[Thinkscape]

I second @DASPRiD - that's exactly why ErrorHandler was created, so I see no reason not to use.

[danizord]

Prefixing it with @ is not enough?

[DASPRiD]

That's not allowed within ZF2.

[Maks3w]

There is another approach. Just silent the warning in the unit test and let it throw in production.

[Ocramius]

@Maks3w I'd go with the fix in the test itself (expect a warning)

@ezimuel any suggestions?

[Maks3w]

ping @zendframework/community-review-team

[jkavalik]

should be possible to check for warning thrown in phpunit (?convert to exception) for php 5.6+

[Maks3w]

As I said. false is the expected return type instead of throw exceptions.

[jkavalik]

@Maks3w I meant only in unit test - split them to good X bad and for bad ones something like

 if(phpver >= 5.6) phpunit.expect-warning(...);

or

if(phpver >= 5.6) {
    phpunit.warning-to-exception = true;
    phpunit.expect-exception(phpunit-warning-exception);
}

[Ocramius]

@ezimuel what is your take on this? Is a warning expected?

[ezimuel]

I think the fix proposed by @Maks3w is ok in that specific case. The use of Zend\Stdlib\ErrorHandler seems to be verbose here. False is perfectly fine because it evidences that the number that you are trying to create with BigInteger is not valid. We don't need more info on that.

[Ocramius]

@ezimuel can you merge then? Or eventually I can do that as well.

[ezimuel]

@Ocramius go for it, thanks!