Open j1mbl3s opened 4 years ago
Watched bills with changes could be displayed on this schedule while another, less frequent schedule could display the rest of the results, with the scan
/query
command displaying the results as needed.
in general we track about 180,000 bills and have an average 3 hour lag between a state publishing any random update and the system picking it up.
State schedule vary wildly from near real-time to once a day, and the Timmy the Intern factor in data entry delay from Clerk staff.
And as it relates to public service Pull API calls there is a 1 hour cache on any specific set of parameters, though 1-2 checks per day is sufficient for most common identification/update approaches
– Sean, from Legiscan
My argument for not locally caching bills is as follows: The scan
command, which does the API queries, is run very infrequently (one to two times a day). It is not a common command to use. The other commands (ignore/watchlist/etc.) operate on the database.json/watchlist, which acts like a cache.
The API documentation says that, within a 1-hour schedule for
search
/searchRaw
, it will serve unchanged cache data (see page 7).