ikxeno commented on PR #13276:
URL: https://github.com/apache/maven/pull/13276#issuecomment-5891628016

   @gnodet 
   
   Default is `--mode effective` now, `--mode raw` the opt-out.
   
   Effective runs the raw pass first. `build()` with `BUILD_EFFECTIVE` says 
nothing
   about the reactor, so on its own it passed a project whose `<modules>` 
points at
   a directory that isn't there. No second read, the file and raw models are 
cached
   per source (`DefaultModelBuilder:1812`, `:2168`).
   
   `--local-repository`, `--temp-local-repository` and `maven.repo.local` all 
work.
   I set the path with `Session.withLocalRepository` on the session that comes 
back
   rather than passing it to `ApiRunner.createSession`, because two things can 
drop
   that argument: `settings.getLocalRepository()` wins over it at 
`ApiRunner:594`,
   and the injector can hand back another tool's pl:333`
   binds a qualified `@Provides` under the unqualified key as well, so mvnup's 
can
   answer `getInstance(Session.class)`. Neither f is what
   made it worth pinning. Same reason I re-derive the remote repositories from 
the
   settings. File that injector one separately?
   
   The shared session I undid. It makes the verdir:
   derived sessions share `mappedSources`, so a GA read from an earlier POM 
answers
   a later one's parent lookup. Same two files, oan error
   about the wrong parent. There's a test now. Every pass walks the reactor 
from the
   root again, so sharing saved no reading either
   
   Your javadoc sketch has `RequestType` controllnever
   reads it. I wrote what's true instead. Want it to actually dispatch?


-- 
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: [email protected]

For queries about this service, please contact Infrastructure at:
[email protected]

Reply via email to