great contributions <@U01MV4U2EV9>!
<https://gith...
# generalj
João Pedro
03/31/2022, 1:45 AMgreat contributions @thdxr!
https://github.com/sensedeep/dynamodb-onetable/pull/308
João Pedro
03/31/2022, 1:46 AMI’ve been using that library as well - it seems like the most mature of all currently - but I still feel like the typescript support could definitely be better, so good on you for contributing!
t
thdxr
03/31/2022, 1:50 AMhaha thanks! I didn't finish though - I started trying out electrodb which is extremely typescript oriented
thdxr
03/31/2022, 1:50 AMit's pretty amazing
j
Jono Allen
03/31/2022, 2:02 AMive been using sst and one table aswel! Currently stuck on how to do pagination…
a
Adrián Mouly
03/31/2022, 10:03 AMWhats the idea behind using one-table concept?
Adrián Mouly
03/31/2022, 10:03 AMDoesnt make things more complicated?
t
thdxr
03/31/2022, 2:06 PMdo you mean this library or the general concept of single table design?
d
Derek Kershner
03/31/2022, 2:12 PMThe What, Why, and When of Single-Table Design with DynamoDB (alexdebrie.com)
Derek Kershner
03/31/2022, 2:13 PMBasically, Single table design is probably the way forward unless you are using GraphQL or you have a pattern that necessitates a Scan.
Derek Kershner
03/31/2022, 2:14 PM(I assumed the latter to Dax’s question)
g
gio
04/04/2022, 9:44 AM@Derek Kershner what do you mean when you say “pattern that necessitates scan”? You can always declare a sparse index on single table
d
Derek Kershner
04/04/2022, 1:44 PMI’m not sure I can pick out every situation, but it seems like it would mostly be for things that are write heavy and rarely scanned. Basically, when the cost of the GSI would outweigh the cost of scanning, and you dont have any querying concerns, it would be more efficient to spin up a second table.
There are also partitioning performance reasons, like when that GSI would cause a single partition to form and you need more than that many reads or writes per second.