Strict XML parser breaks Maven backward compatibility

[cstamas]

This change:
761ac42

The p-u XML parser is made stricter, and adhere standard, which is fine, but there is a problem: older Maven versions were able to parse, hence build and deploy projects that used POMs that does not adhere to strict(er) XML parsing. If it happens for user to pick up such POM (ie. among transitive dependencies), there is no breakout for user from that situation.

Also, see issue https://issues.apache.org/jira/browse/MNG-7697

Proposal: add some "lax-standards" switch for this. I agree that project POMs should fully adhere to XML 1.0 (so MNG-7697 is IMHO a fluke), but dependency POMs should be parsed with "laxed" standards, just like dependency POMs validation is laxed as well, given they are probably built and deployed with Maven versions pre-3.9.0, and MAY contain invalid characters as those versions would allow emojis in comments...

ping @belingueres

[michael-o]

I imagine a system property for this...

[cstamas]

Why a system property? Am fine to keep strict XML 1.0 parsing for project POM, but lax parsing for dependency POMs... how would a system property help here?

[michael-o]

Why a system property? Am fine to keep strict XML 1.0 parsing for project POM, but lax parsing for dependency POMs... how would a system property help here?

How is Plexus supposed to make the difference?

[slawekjaranowski]

can be couse of mojohaus/build-helper-maven-plugin#168

[belingueres]

Hi:
Sorry for the late reply.
@cstamas : I confirm that change is detecting the emoji as an invalid character.
@slawekjaranowski: the mojohaus/build-helper-maven-plugin#168 I confirm this is a bug. I'll be working on this issue first.

[gnodet]

Hi:
Sorry for the late reply.
@cstamas : I confirm that change is detecting the emoji as an invalid character.

Fixed with #240

@slawekjaranowski: the mojohaus/build-helper-maven-plugin#168 I confirm this is a bug. I'll be working on this issue first.

Fixed with takari/polyglot-maven#259 / eclipse-tycho/tycho#2176 . I don't think this is an issue for the mx parser.

@belingueres fwiw