High level new features and changes:

• (change) filtering is enabled by default
• (new) filtering can be suppressed on per-repository basis by appending configuration key with .repoId=false.
• (change) both filters enhanced to become really usable.

GroupId filter

Is enabled by default (but remains dormant if user does not provides input). It becomes active only when user provides input for it.

Input vastly improved and now supports modifiers as well (Resolver 1.x modus operandi was =org.apache ONLY). Hence, upgrading from Resolver 1.x to Resolver 2.x leads to "broadened" filtering but that is fine is was user usually wants (Resolver 1.x had to enumerate all Gs for nested G, ie org.apache.maven, org.apache.maven.plugins etc... same effect with Resolver 2.x is one liner: org.apache.maven (as it means now "and below"). Added support for negation (!) and limiter (=). Example groupId filter file:

# My file                   (1)
                            (2)
org.apache.maven            (3)
!=org.apache.maven.foo      (4)
!org.apache.maven.indexer   (5)
=org.apache.bar             (6)

Lines 1, 2 ignored. Line 3 "allow org.apache.maven and below". Line 4 "not allow org.apache.maven.foo only". Line 5 "not allow org.apache.maven.indexer and below". Line 6 "allow org.apache.bar only".

Prefix filter

Is enabled by default and improved to try to auto-discover prefix file published by remote repository. If prefix file is discovered, common resolver means are used to resolve and cache the prefix file in user local repository. If discovery fails, user provided prefix file is attempted (and used, if found). Finally, if no input, filter will not interfere. Order of precedence:

• user provided prefix file (if found, will be used, no discovery happens)
• auto discovery

Caching is now possible with fix done in #1491