nielsbasjes commented on pull request #26:
URL: https://github.com/apache/maven-shade-plugin/pull/26#issuecomment-873648803
@michael-o
1. Yes, this is indeed a hard problem with a lot of side effects to think
about.
2. I already did that check (as I mentioned in the description) with both
3.8.1 and a maven that includes my patch; Note that it also passed the CI build.
3. I'll leave this open for now.
I read in the comments that someone had a single project with over 5000
modules which in my experience sounds very extreme.
Perhaps a monorepo that is also a single build?
I checked the mentioned Apache Camel (586 pom.xml files) and some projects
which I expected to have a large number of modules (just counted the pom.xml
files): Flink (179), Hadoop (113), Nifi (592) and Spark (36)
So my main question is to what extend should maven support so many modules?
Even for very large projects having > 1000 modules seems like "too many".
I'll see what I can do at the maven end to make a test for this.
1) I read about the algorithms in maven that determine the actual build
ordering that are affected by this change.
At this point I do not have enough knowledge about this to come up with
a test that verifies this behavior to remain correct.
2) As far as I can tell the only way is to add something that changes the
reactor build ... like the test of this pull request.
Simply migrating this (now failing) test to maven would make the
maven-shade-plugin a part of the tests in the maven build.
Is this ok ? Or do you have a better proposal on how to reproduce this?
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To unsubscribe, e-mail: issues-unsubscr...@maven.apache.org
For queries about this service, please contact Infrastructure at:
us...@infra.apache.org
