Open yarkinwho opened 8 months ago
In 5.0, the serialization of get_table_rows
is off the main thread. See https://github.com/AntelopeIO/leap/pull/1054 . Also the time constraints are different in 5.0, see https://github.com/AntelopeIO/leap/pull/1139 and get_table_rows
calls are run in read-only mode meaning it can use the full read-only-threads
setting of nodeos
to process them in parallel.
You can safely increase the default request from 10 rows to a much larger number.
Currently when the frontend trys to fetch table rows from EOS, it will fetch v1/chain/get-table-rows in one shot instead of doing any paging.
The result will be limited to 10 rows by default. It is not enough if the FE want to read things like the ERC token list. Currently the limit is set to 20 in the query and can be further relaxed without caused major trouble given the amount of data per row is quite small.
In the future we should optimize the way we query the rows.
Note that current code can work for quite a while as long as we do not open the registration of tokens to public. So it's fine to ignore this issue in this case. But we should solve this issue when we are doing the rework of the FE for opening the registration to public.