I'm getting errors about being "unable to index remote maven repositories" since upgrading; the repos in question are private S3 repos that we deploy our own jars to via lein-essthree

Same issue with s3wagon private

I think those simple wagons don't build global indexes of their contents like the real maven repo servers do

They also have locking problems

Yeah, was trying to follow these instructions to index manually, but it didn't make a difference: http://stackoverflow.com/questions/5349261/how-to-index-a-maven-repo-without-nexus-artifactory-etc

...those are pretty old though

Gonna try that on our repos later

Well, I tried generating an index with https://github.com/apache/maven-indexer 5.1.2 via https://mvnrepository.com/artifact/org.apache.maven.indexer/indexer-cli/5.1.2-4333789

...but I still see an error in IntelliJ under the "repositories" tab

"Wagon exception connecting to" my repo

I'll open a ticket for it

Or add to the existing one actually : https://github.com/cursive-ide/cursive/issues/1064

I don't have this issue, I use s3-wagon-private and cursive can download those deps just fine

Hmm, yeah, I can download them OK but not index them, seemingly

...though maybe intellij is just getting them from my local repo, now that I think about it

@timgilbert how did you mount the s3 as a drive so the indexer could run on it?

I didn't. What I'm doing is deploying datomic-pro jars out to S3, so I've got a little shell script that installs it to a local directory via maven and then syncs that to S3

Most of it is this little block:

tmp=$(mktemp -d -t 'datomic-jar-deploy')
mvn deploy:deploy-file \
    -Dfile=${jar} \
    -DgroupId=com.datomic \
    -Dversion=${version} \
    -DartifactId=datomic-pro \
    -Dpackaging=jar \
    -Dmaven.repo.local=${tmp} \
    -DpomFile=pom.xml \
    -DcreateChecksum \
    -Durl=file:${tmp}

runMavenIndex $tmp

aws s3 sync ${tmp} ${S3_MAVEN_ROOT} \
    || die "aws s3 sync ${tmp} ${S3_MAVEN_ROOT} failed: $?"

...plus this to do the indexing:

function runMavenIndex() {
  local repo=$1
  java -jar ${INDEXER_JAR} -r ${repo} -i ${repo}.index -n "my-repo-name"
}

don't you need to run the index against the entire contents of s3, not just the particular artifact you are uploading?

although I would still expect an index to work, just only have one entry in it

Well, yeah in theory, but I've only got a few other jars I care about so I wanted to get this working first

I could see it also working as a "download full s3 repo; index locally; sync back to s3" sequence, not super efficient but presumably workable

But without IntelliJ getting the index file to start with that seems moot

Is there any integration with eastwood? Some like "eastwood this ns"?