#75 fixed broken validation of page_size parameter

[ppaulis]

Signed-off-by: Pascal Paulis ppaulis@gmail.com

Q	A
Documentation	no
Bugfix	yes
BC Break	no
New Feature	no
RFC	no
QA	no

Description

This fixes the bug described in #75 .

The concerned part is:

if (is_numeric($value) && intval($value) !== $value) { return false; }

This validation makes no sense the way it is, because the GUI sends strings like "50", and in this case intval($value) !== $value always evaluates to true, due to the strict comparison. Instead, the validation should be :

if (!is_numeric($value) || intval($value) != $value) {
    return false;
}

as proposed in this PR.

The Unit Test in this case was incorrect because it only tested on integers.

[ppaulis]

@froschdesign Here's the new PR with the commit correctly signed off and the fixed Unit Test.

[froschdesign]

@ppaulis

The Unit Test in this case was incorrect because it only tested on integers.

But this means that integers are valid and we need a test for this. Therefore, a new dataset needs to be added to the provider instead of changing existing ones.

[ppaulis]

@froschdesign I completed the tests, so that now all kinds of values are tested. I reorganized the code a little bit to avoid duplicated code fragments.

Here's a resume:	value	return
'25'	valid
'-1'	valid
25	valid
-1	valid
'string'	not valid
'25.5'	not valid
'-2'	not valid
25.5	not valid
-2	not valid

[weierophinney]

The CS and Psalm failures are due to an invalid lockfile; I'll fix that in a different PR and then we can rebase against that here.