Publication does not contain any dependencies in publication

[andriihorpenko]

Description

Since #198, MDG seems to not copy dependencies from the original config into the newly created reobfConfig.
I assume the issue lies where MDG replaces original config with the new config, which adds reobfuscated artifact only.

ModDevGradle/src/legacy/java/net/neoforged/moddevgradle/legacyforge/dsl/ObfuscationExtension.java

Lines 146 to 147 in 5b0a9bf

	java.withVariantsFromConfiguration(config, ConfigurationVariantDetails::skip);
	java.addVariantsFromConfiguration(reobfConfig, spec -> {});

Reproduction

Add any runtimeOnly dependency to both legacytest and testproject projects and publish them.

• testproject will have said dependency listed in POM and Gradle Module Metadata
• legacytest will not have it

Notes

See this Discord thread for more info.

[Icosider]

As far as I know, MDG avoids any transitive dependencies, as they cause unpredictable bugs. Example: duplicate dependencies

[shartte]

No this is indeed not intentional Yuri ***@***.***> schrieb am Mi., 22. Jan. 2025, 12:26:

…

As far as I know, MDG avoids any transitive dependencies, as they cause unpredictable bugs. Example: duplicate dependencies — Reply to this email directly, view it on GitHub <#227 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AAJT6V7X4OCYROPG5YODOOD2L555TAVCNFSM6AAAAABVUUORQOVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMZDMMBWHE4DIMRUG4> . You are receiving this because you are subscribed to this thread.Message ID: ***@***.***>

[Icosider]

@shartte, the most interesting thing is that this problem is only in Legacy, as well as the problem associated with publishing multiple builds. Is MDG very strict and closed in such things for some reason, compared to other functionality, or are these just unaccounted-for likely test outcomes?

[shartte]

@shartte, the most interesting thing is that this problem is only in Legacy, as well as the problem associated with publishing multiple builds. Is MDG very strict and closed in such things for some reason, compared to other functionality, or are these just unaccounted-for likely test outcomes?

That's not really interesting. MDG only has to mess with jar or publications because it has to replace the jar for reobfuscation.