Closed valentinedwv closed 1 year ago
cleaning up and documenting it more now, will get in shared place soon, linked from this issue
starting diagram / documentation of summary workflow https://github.com/MBcode/ec/blob/master/summary.md
The 1st part of getting a crawl uses a workaround to get quads/repo for the summarization
So, we need to add a note not to disable to milled:true to the doco
I was thinking about disabling the default templates for glcon configuration no longer run milled gleaner, because Nabu loads the data to the graph.
btw the fix_runX.sh process can take summoned as well, it just takes longer; then it makes the quads, and now materialize_summary.py takes the repo.nq to repo.tll for uploading. Just for safety sake, I think it is best to upload both of these files, to make sure the graph-urns are exactly the same btwn them. The latest PR gets this all going w/blaze vs fuseki, and later we still have the option of query of file via lib vs using any endpoint at all
specifics: