Our estimation of the number of records is puzzling users
Currently we're setting in the client the maximum of records to (display_limit * 100)+
when there is more than this number of records for the search of the user.
Some users do not understand this because they know there is much more records than (display_limit * 100)
.
But on the other hand we can not just send the maximum number of records as it would puzzle them also if they refine the search and this maximum number is just the same as before.
Cédric suggested using (nb_records - limit) / 2 + limit
when there is a record which is somewhere in the middle.