Feat/port lotus dm selector retrieval 6393 to estuary #5
Conversation
replication.go
Outdated
| prog.endErr = err | ||
| return err | ||
| } | ||
|
|
||
| // FIXME need to do something with this | ||
| panic(contentRoot) |
There was a problem hiding this comment.
When you do a partial retrieval - you get a new CID for your sub-dag. The client doesn't know what it is, just the path to it. The code pulls it out for you by re-running the selector, and then you do... something with it.
There was a problem hiding this comment.
@whyrusleeping an alternative would be to run the selector-over-your BS lazily, when the user requests their data back. Then changing the functions args won't be needed.
It is currently done this way to minimize the amount of time for GC ( since you can't pin the partial dag )...
Dunno - thoughts welcome.
ribasushi
force-pushed
the
feat/port_lotus-dm-selector-retrieval_6393_to_estuary
branch
from
699ea2f
to
c3292cb
Compare
|
@elijaharita that last commit wires in c3292cb#diff-101af5442d7541ffa58c8cda4f725ff032482209abe78f90545af14996457020R347-R350 You can use it like so: filecoin-project/lotus#6393 (comment) |
…selector-retrieval_6393_to_estuary
…selector-retrieval_6393_to_estuary
|
@whyrusleeping this bad boy is brought up to master again. Do you want to merge it, |
|
I will keep this out until we land the graphsync upgrade in lotus, to minimize churn. Will reopen in couple weeks when the time is right. |
…eat/graphsync-opts update some memory limits for the graphsync client
@whyrusleeping needs more work on the edges, but you should be able to get it from here