World Conquest doesn't properly pass WML schema validation

[Pentarctagon]

Currently the EDITOR scenario created for WC is hidden from WML schema checks behind and #ifndef:

wesnoth/data/campaigns/World_Conquest/scenarios/WC_II_scenario.cfg

Lines 242 to 244 in 545e5ff

	#ifndef SCHEMA_VALIDATION
	{WC2_SCENARIO_NEW multiplayer WC_II {WC_II_CAMPAIGN_NAME_3P} 0}
	#endif

however the WML schema should be properly updated so that this isn't needed. The issue is that when running the EDITOR validation, it errors saying that nplayers isn't a valid attribute to be present. I assume most likely this is caused by the editor using [multiplayer] rather than [scenario]. This editor specific WC scenario was added in 8daad7d.

[gfgtdf]

The [generator] also contains the line nplayers={PLAYERS} twice

[CelticMinstrel]

The double attribute is wrong but would not be the cause of a schema validation failure – the first instance of the attribute will be overwritten by the second before validation runs, I believe.

I think basically what is required is a schema entry for the WC map generator?

[gfgtdf]

the first instance of the attribute will be overwritten by the second before validation runs, I believe

Here you said something different though:

24cf433

[CelticMinstrel]

The two things are not related.

WML doesn't support having two attributes with the same name. If it happens anyway, one of them will be ignored at parse time. (I guess ideally that would be an error but I don't think it is.)

The linked commit notes that the order of attributes can make a difference to the schema validator. If attribute A is only valid when attribute B is present, then it will fail to validate if attribute A appears before attribute B but pass if attribute B appears before attribute A. (I guess this should be considered a bug in the schema validator, but I'm not really sure how to fix it.)

Making WC validate just means we need to add an if or switch to the schema definition of [generator] that contains any special keys supported by the WC map generator. Or alternatively, specify that any key is allowed.

[gfgtdf]

The two things are not related. ... The linked commit notes that the order of attributes can make a difference to the schema validator

And this issue is about only the schema validator, or am i missing something ?

[CelticMinstrel]

I think this issue would be about the schema, rather than the schema validator?