We managed to speed up getRepo a ton and reduce its memory footprint as part of #88, but it's still too expensive due to the heavy use of datastore lookups, one per block. It was costing us almost $500/day in datastore reads alone over the last two days.
So sadly we have it disabled right now, but we need to find a way to make it less expensive and turn it back on sooner or later.
We managed to speed up
getRepoa ton and reduce its memory footprint as part of #88, but it's still too expensive due to the heavy use of datastore lookups, one per block. It was costing us almost $500/day in datastore reads alone over the last two days.So sadly we have it disabled right now, but we need to find a way to make it less expensive and turn it back on sooner or later.