fix: cluster-id 0 discv5 issue #2562
Conversation
| let clusterZeroConf = ClusterConf.ClusterZeroConf() | ||
| conf.pubsubTopics = clusterZeroConf.pubsubTopics | ||
| # TODO: Write some template to "merge" the configs | ||
| # cluster-id=1 (aka The Waku Network) |
There was a problem hiding this comment.
no real change from here. just idendation.
|
You can find the image built from this PR at Built from 12f693a |
alrevuelta
changed the title
alrevuelta
requested review from
rymnc,
chaitanyaprem,
jm-clius,
stubbsta and
Ivansete-status
alrevuelta
force-pushed
the
fix-cluster-zero
branch
from
707e0ca to
4f7a4ea
Compare
| conf.discv5BootstrapNodes & twnClusterConf.discv5BootstrapNodes | ||
| conf.rlnEpochSizeSec = twnClusterConf.rlnEpochSizeSec | ||
| conf.rlnRelayUserMessageLimit = twnClusterConf.rlnRelayUserMessageLimit | ||
| discard |
There was a problem hiding this comment.
We discard at this point because we are not willing to set any particular configuration when either cluster_id == 0 or 1, right?
There was a problem hiding this comment.
correct. anyone can define their own clusterId eg =999 and there is no config we set in there.
| of 0: | ||
| let clusterZeroConf = ClusterConf.ClusterZeroConf() | ||
| conf.pubsubTopics = clusterZeroConf.pubsubTopics | ||
| # TODO: Write some template to "merge" the configs |
There was a problem hiding this comment.
What do you mean by "merge the configs" ?
There was a problem hiding this comment.
imagine
type A:
field1
field2
typeB:
field1
field2
field3
field4
which is similar to WakuConf and ClusterZeroConf. Note that typeB is a superset of typeA.
Right now we are "merging" boh manually.
let a = TypeA(...)
let b = TypeB(...)
# we want a config to override b
b.field1 = a.field1
b.field2 = a.field2
this is i) error prone, ii) if number of parameters increase, we have to assign them manually.
so I'm suggesting with some templating, writing a merge() function so that.
let a = TypeA(...)
let b = TypeB(...)
merge(a, b)
or something like this. with this we no longer need all this
Description
topicsbut introduced a regression.topicsflag default value/waku/2/default-waku/protowas being used when constructing the ENR, after removing that flag, we no longer had a default value.wakunode2 --cluster-id=0 ...discv5 was not working properly, since the node had no default pubsub topic configured.cluster-id=0.pubsub-topic.