Fix hashing of asset directories

[rcannood]

Files.walkFileTree does not guarantee any specific order for iterating over files within a directory:

Considerations When Creating a FileVisitor

A file tree is walked depth first, but you cannot make any assumptions about the iteration order that subdirectories are visited.

-- source

This causes the sha of HashBuilder.hashDirSha256 to be different in certain situations (#6112).

This PR fixes the issue by storing all of the (path, sha256) in an array list, sorting by path, and then passing them to the hasher.

Notes:

• The code in preVisitDirectory seems like it would cause a null pointer exception if the condition (base != null) is false -- so maybe base is always not null? What should happen in this code chuck if base does happen to be null?
• It's hard to add a unit test for this, since the issue only arises in very specific situations.

[netlify]

✅ Deploy Preview for nextflow-docs-staging canceled.

Name	Link
🔨 Latest commit	31e1a8a
🔍 Latest deploy log	https://app.netlify.com/projects/nextflow-docs-staging/deploys/68430ca527377e0008af5881

[rcannood]

Hi everybody! Can I get some feedback from the Seqera team on this issue + PR? 🙇

[pditommaso]

Thanks for reporting this, very tricky. Let a comment to improve the implementation

[rcannood]

Thanks for the feedback, @pditommaso ! I swapped the ArrayList for a TreeMap, which makes more sense.

Regarding the potential NullPointerException in preVisitDirectory, which solution would you prefer?

[bentsherman]

I think this function can be simplified by simply applying hashUnorderedCollection() to the set of key-value pairs. I will give it a try...

[bentsherman]

@rcannood let me know if my changes work for you

[rcannood]

@rcannood let me know if my changes work for you

Thanks @bentsherman ! I'll run it on my other partition when I'm back at my PC.