#011

• make sure the conclusions ties back into the new introduction,
• make sure all the further work is in the further work chapter,
• rewrite the abstract to include the motivation and relevance of the work.

I wrote a script to generate query strings for our finders (like the AAIB reports finder), so we can load test them. Most finders have way too many (or infinitely many) possible queries to test them all, so it generates a random sample. Here are some examples:

• {"aircraft_category"=>["sport-aviation-and-balloons", "commercial-fixed-wing", "general-aviation-rotorcraft", "unmanned-aircraft-systems", "commercial-rotorcraft"], "report_type"=>["correspondence-investigation"], "date_of_occurrence[from]"=>"30-4-31", "date_of_occurrence[to]"=>"57-4-25"}
• {"aircraft_category"=>["general-aviation-rotorcraft", "sport-aviation-and-balloons", "commercial-rotorcraft"], "report_type"=>["safety-study", "pre-1997-monthly-report", "formal-report", "annual-safety-report", "field-investigation"], "date_of_occurrence[from]"=>"21-10-16", "date_of_occurrence[to]"=>"15-8-24"}
• {"aircraft_category"=>["commercial-fixed-wing"], "report_type"=>["pre-1997-monthly-report", "special-bulletin", "foreign-report", "annual-safety-report"], "date_of_occurrence[from]"=>"30-6-8", "date_of_occurrence[to]"=>"69-2-12"}
• {"aircraft_category"=>["general-aviation-fixed-wing", "sport-aviation-and-balloons", "unmanned-aircraft-systems"], "report_type"=>["annual-safety-report"], "date_of_occurrence[from]"=>"33-8-29", "date_of_occurrence[to]"=>"61-12-15"}
• {"aircraft_category"=>["commercial-rotorcraft"], "report_type"=>["field-investigation", "pre-1997-monthly-report", "foreign-report", "special-bulletin", "annual-safety-report", "safety-study"], "date_of_occurrence[from]"=>"71-5-24", "date_of_occurrence[to]"=>"99-6-30"}
• {"aircraft_category"=>["commercial-fixed-wing"], "report_type"=>["field-investigation"], "date_of_occurrence[from]"=>"81-2-15", "date_of_occurrence[to]"=>"7-2-11"}
• {"aircraft_category"=>["general-aviation-fixed-wing"], "report_type"=>["foreign-report"], "date_of_occurrence[from]"=>"4-5-31", "date_of_occurrence[to]"=>"2-1-7"}
• {"aircraft_category"=>["commercial-fixed-wing"], "report_type"=>["correspondence-investigation", "pre-1997-monthly-report", "special-bulletin", "safety-study"], "date_of_occurrence[from]"=>"67-11-27", "date_of_occurrence[to]"=>"82-8-8"}
• {"aircraft_category"=>["unmanned-aircraft-systems"], "report_type"=>["formal-report"], "date_of_occurrence[from]"=>"1-9-2", "date_of_occurrence[to]"=>"37-3-25"}
• {"aircraft_category"=>["general-aviation-rotorcraft", "commercial-rotorcraft", "general-aviation-fixed-wing", "commercial-fixed-wing"], "report_type"=>["correspondence-investigation"], "date_of_occurrence[from]"=>"59-10-5", "date_of_occurrence[to]"=>"23-1-7"}

I finished up some work from a few weeks ago by removing an unnecessary field from link expansion. Unfortunately it’s not a very big win. The most significant change in this series was the first one we did, shrinking organisation links.

On Monday I watched the InSight probe landing on Mars.

dejafu got a small flurry of interest, and is on its way back to Stackage. It seems there’s been some interest in using dejafu at DFINITY, a blockchain startup. One day I’ll figure out why there are so many Haskell blockchain startups—or even why there are any blockchain startups at all.

I started playing Sunless Skies, a game by the Fallen London people in early access. You play the captain of a locomotive exploring the skies for Queen and Empire. It’s pretty fun, and hard; but fortunately when you die and start over again, you inherit a bit of what your last captain had, so it gradually gets easier.