Limit path flatten max length

[rvesse]

When TransformPathFlattern/TransformPathFlattenAlgebra run they expand some paths of the form :x :p{N,M} ?y into multiple triple patterns within a BGP e.g. :x :p ?q0 . ?q0 :p ?y etc. This is safe for short values of M but can result in extremely large expansions if M is a large value. The large expansion is unlikely to make the query run any faster. This commit introduces a new static control for the maximum expansion length capped at 10 and does not apply full expansion to anything longer than that.

---

• Tests are included.
• Documentation change and updates are provided for the Apache Jena website
• Commits have been squashed to remove intermediate development commit messages.
• Key commit messages start with the issue number (GH-xxxx)

By submitting this pull request, I acknowledge that I am making a contribution to the Apache Software Foundation under the terms and conditions of the Contributor's Agreement.

---

See the Apache Jena "Contributing" guide.

[afs]

I agree with the intent.

just checking here but I think you mean N, not M.
:p{N,M}, it splits into :p{N} / :p{0,M-N} ((pathN N :p) and (mod 0 M-N :p)).
and the first part pathN is written triple patterns.

Unless there is evidence to the contrary, maybe the limit (<10) should be lower - e.g. 5.

[rvesse]

I agree with the intent.

just checking here but I think you mean N, not M. :p{N,M}, it splits into :p{N} / :p{0,M-N} ((pathN N :p) and (mod 0 M-N :p)). and the first part pathN is written triple patterns.

Yes sorry, got the terminology confused

Unless there is evidence to the contrary, maybe the limit (<10) should be lower - e.g. 5.

Yeah I think that's probably sensible, can adjust the PR accordingly