Migrate ext/soap resources to objects

[kocsismate]

Part of php/php-tasks#6. Migrates 2 out of 3 resource types in ext/soap to opaque objects.

[ndossche]

There's two simple failures due to different object numberings, that should be easy to fix.
Is it also possible to split the commit into 2: one for each resource? That would make reviewing easier.
ext-soap test coverage is so-so given that the extension contains complex interactions, so I'm a bit worried about the complexity of such a transition, but we'll see I guess.

[kocsismate]

There's two simple failures due to different object numberings, that should be easy to fix.

Yes, these are fixed locally. I'll look into the memory leak related failure tomorrow. :)

Is it also possible to split the commit into 2: one for each resource? That would make reviewing easier.

I may try it, sure.

[Girgias]

Is it possible to make the change to a generic namespace in the stubs in a separate commit too? As it does make the diff somewhat confusing.

Didn't really have a look at the actual SOAP changes yet as I don't really know how SOAP works

[Girgias]

Nit: EOL before EOF

[kocsismate]

Is it possible to make the change to a generic namespace in the stubs in a separate commit too? As it does make the diff somewhat confusing.

@Girgias Please use the "hide whitespace changes" feature to fix it ^^ But anyway, since I'm currently rewriting the history completely, I'll commit this separately.

[ndossche]

Looks good, tested it as well, just some questions.

[ndossche]

I first didn't understand what this last sentence meant, but I guess you mean that you have to take into account it can be a null value, so you can't pass it to is_resource? However, is_resource will always return false for Soap\Url, so I'm not sure what you mean.

[kocsismate]

This is the upgrading note template we used in the past for all the other conversions, but I'm happy to change it in the name of clarity ^^ It is intended to mean that when checking whether the property "has a value", then previously one may have used is_resource($client->httpurl), but obviously this won't work anymore, so it should be changed to $client->httpurl !== null. Of course, it's mostly a theoretical problem IMO, since the property is private.

[ndossche]

Okay I get it now, thx for explaining.
I think this is unclear indeed, because "value checks" is quite vague. is_resource always checks a value, so writing "value checks using is_resource()" was very weird to me.

[kocsismate]

I removed the "value" word, and added an example. I think that's the most clear we can do. I'll merge my changes, but if the note is still unclear, we can amend it separately.

[ndossche]

Should this be final? Intuitively I'd think so.

[kocsismate]

Yes, I forgot about the final modifier, but it's a good idea!

[ndossche]

Same remark as earlier about the last sentence.

[ndossche]

Should this be final?

[ndossche]

LGTM, thanks.
For wording in UPGRADING, well I'm not sure what's best. I commented on your reply.

[TimWolla]

Suggested change

	typedef struct {
	php_url *url;
	zend_object std;
	} soap_url_object;
	typedef struct soap_url_object {
	php_url *url;
	zend_object std;
	} soap_url_object;

Can you please add a name to the struct, instead of typedefing an anonymous struct? This makes the output of tools like "pahole" easier to understand. The same remark applies further below and possibly in other files as well.

[kocsismate]

Yes, sure!