Myroslav

It turns out that the layers with their IPFS references are already [cached](https://github.com/propensive/fury/pull/1173/files#diff-d269152254dde0ab8cdda98a32d82707R132) in memory since PR #1173. However, the logs still show a lot of repeating messages "Checking for...

The publish command of Coursier is considered experimental. It is disabled by default in the command line application, and it seems to be entirely missing from the artifacts published at...

The threads named "FileSystemWatchService" are created by the OpenJDK implementation of [file system poller](https://github.com/openjdk/jdk/blob/master/src/java.base/share/classes/sun/nio/fs/AbstractPoller.java#L64). I don't think the directory-watcher library has any control over them. Since Fury creates one directory...

I don't think that would be a useful addition to the interface. In my opinion, two short commands are more convenient than a single long one, as long as they...

If we ever have the interactive prompts, they should be enabled by default, otherwise the user might never learn that they exist. It would be nice if they could be...

Another possible option is creating a separate command that would guide the user through every decision needed to set up a layer from scratch, such as creating projects and modules,...

The problem is also present when Fury commands are run in the standalone mode. This means that the problem lies either within Bloop, or in the way Fury communicates with...

The problem does not appear if Bloop is killed after step 4 (that is, after the failed compilation).

The commit https://github.com/scalacenter/bloop/commit/92c23aefdaf2602a6ffeb357518ef4a6db445d0f might be related.

This issue is much less visible after PR #949 has changed the way BSP connections are managed, but the root cause (in my opinion, that's the strange behaviour of Bloop)...