maven-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Jörg Schaible <>
Subject M3 regression: Unreliable results, because reactor fails to calculate proper build order
Date Mon, 30 Jan 2012 13:57:32 GMT
Hello Maven devs,

although we use M3 now for quite some time, we finally had to recognize, 
that the results are completely unreliable, because the reactor fails to 
calculate the proper build order. I've created a test case in MNG-5207. It 
happens when transitive dependencies are overwritten in the 
dependencyManagement section with SNAPSHOTs that are part of the same 
reactor build. In such a case the reactor of M3 fails to calculate the 
proper order.

When the local repository is cleared and no SNAPSHOTs are available, the 
reactor build simply fails at one point, because the required SNAPSHOTs are 
not yet available. The situation gets worse when the project has been built 
once, then our ZIPs, EARs and WARs may silently contain stale SNAPSHOT 
versions and the result of the build is simply rubbish.

This behavior is an absolute show stopper for us and forced us to ban M3 
completely from development, because we cannot afford unreliable results. 
M221 has absolutely no problems in this regard, it calculates the proper 
build order everytime. All tested M3 versions (3.0, 3.0.1, 3.0.2, 3.0.3, 
3.0.4) fail poorly. Demonstrated with MNG-5702, M221 succeeds, any M3 
version fails.


To unsubscribe, e-mail:
For additional commands, e-mail:

View raw message