...
hydra-derivatives has some hoops you have to jump through to get build to pass
Dependencies on tools used to generate derivatives
One is Kakadu
Some of the files needed to build were missing, had to find them an alternative way
Hydra Derivatives has this as an ongoing problem
Suggestions?
Trey - like integration tests existing, but don't need to deal with this then
In case of Kakadu, we've left it behind and they've been stricter about licensing.
But there is not a great alternative
Chris: In short term want to cut a release of hydra-derivatives
have a PR - dropped testing on Ruby 2.4 b/c EOL and there are issues with the docker images
does dropping testing of Ruby 2.4 require a major version bump? Still works but not getting tested
Trey: No.
Tamsin: strictly speaking it's a nice thing to do but not super important. But sometimes things that are
What is downside to dumping major version?
Chris - if stricter versioning in gemfiles, will have to bump them to get the new version.
Tamsin: Don't feel it's super important either way. Do it how you want.
TOPIC: RDF upgrade
Led by Chris Colvard
Wanted to check with Tamsin on where things are with RDF upgrade.
Concern that I have is that Greg Kallick(?) has a PR into LDP or to 3.2 for all RDF gems
In Avalon-space we're stuck in an older rdf-vocab space as some things have changed
Might drop rdf-vocab long term
Tamsin
What happened over holidays - Greg rewrote in-memory data repository
Past version used a Hash Arraymapped Tree to make small memory footprint while allowing for fast updates and fully-snapshottable transactions
Grab data in memory at any point and just hold it in memory
Downside: data structure was fundamentally unordered - find for RDF but not everything else.
Greg flipped it back - not sure motivation
New implementation that looks a lot like an old implementation
Snapshotting is retained
Some breakages as a part of that - some naming changed. On Monday Tamsin renamed to the previous name. Should be fixed.
However, a little worried about the memory properties of this datastructure
If being conservative, might not want to upgrade RDF to latest yet - especially where people are doing big reindexes, this might have issues
Going to do some benchmarking/testing - we might want to revert if it doesn't hold up
Greg concerned about breaking RDP/active-triples, think these changes are related to that
Chris put some comments but not much in the PR review
Seems like upgrade to 3.2 forced all other RDF gems to 3.2 , concerned about that
Tamsin - we may be able to change that. Will look at it this afternoon.