Alex Klibisz

> All the GitHub App stuff (auth + repos) would then be handled in a separate program that runs before Scala Steward and prepares the askpass program and repos.md. I...

Agreed it would be good to clean up that logging. Are there any other cases where scala-steward can fail and still complete w/ a successful response? I feel like I...

I looked into this a bit more today, running scala-steward on a private repo. I noticed the following interesting behavior: I created an empty project with a build.sbt like this:...

I was able to replicate this behavior in a "real" project. For now I think I'm just going to avoid the `pullRequests.grouping` configuration.

I added some println debugging in my fake circe project, on this commit: https://github.com/scala-steward-org/scala-steward/pull/2907/commits/39b618f79d1c850baba647018614f67b38674a8f. Specifically in the findUpdateReplacements and applyUpdateReplacements methods in EditAlg. Below are the respective logs. I pulled...

The main difference I see is in the update parameter passed to findUpdateReplacements: * without grouping: a single call o findUpdateReplacement w/ `update = ForGroupId(NonEmptyList(ForArtifactId(CrossDependency(NonEmptyList(Dependency(io.circe,ArtifactId(circe-core,Some(circe-core_2.13)),0.13.0,None,None,None))),NonEmptyList(0.14.4),None,None), ForArtifactId(CrossDependency(NonEmptyList(Dependency(io.circe,ArtifactId(circe-parser,Some(circe-parser_2.13)),0.13.0,None,None,None))),NonEmptyList(0.14.4),None,None)))` * with grouping: two...

So there might be two questions here: 1. Why does presence or absence of grouping produce different updates for the same dependency? 2. Why does a `ForArtifactId(CrossDependency(...))` update produce this...

One last though, and then I have to wrap up my investigation soon: I added another println to `applyNewUpdate` in NurtureAlg, to print the `data.update` parameter. * Without grouping: `data.update...

Thanks for the insight. I'll have to re-read a few times to make sure I follow. In the meantime, do you see any way to avoid the incorrect warnings?

Results on cx41 were: |Model|Parameters|Recall|Queries per Second| |---|---|---|---| |eknn-l2lsh|L=100 k=4 w=1024 candidates=500 probes=0|0.379|133.326| |eknn-l2lsh|L=100 k=4 w=1024 candidates=1000 probes=0|0.446|118.878| |eknn-l2lsh|L=100 k=4 w=1024 candidates=500 probes=3|0.635|101.896| |eknn-l2lsh|L=100 k=4 w=1024 candidates=1000 probes=3|0.717|94.297| |eknn-l2lsh|L=100 k=4...