Graph node dev mode #5982
Graph node dev mode #5982
Conversation
incrypto32
force-pushed
the
krishna/graph-dev
branch
from
2783041 to
a414de1
Compare
incrypto32
force-pushed
the
krishna/graph-dev
branch
from
a414de1 to
ff19f99
Compare
incrypto32
force-pushed
the
krishna/file-link-resolver
branch
from
3ef16d5 to
23e91f1
Compare
incrypto32
force-pushed
the
krishna/graph-dev
branch
from
ff19f99 to
e5a9e23
Compare
| file_link_resolver | ||
| } else { | ||
| Arc::new(IpfsResolver::new(ipfs_client, env_vars.cheap_clone())) | ||
| }; |
There was a problem hiding this comment.
match would be more idiomatic here. Alternatively, have run take a Arc<dyn LinkResolver> and leave it up to each caller what they want to pass in
|
|
||
| // Once we receive an event, wait for a short period of time to allow for multiple events to be received | ||
| // This is because running graph build writes multiple files at once | ||
| // Which triggers multiple events, we only need to react to it once |
There was a problem hiding this comment.
It would be more robust to have graph-cli touch a file as the very last thing it does and only watch that. That would also make exclusions etc. unnecessary. For now though, this is fine.
| // If no exclusions, process all events | ||
| if exclusion_set.is_empty() { | ||
| return true; | ||
| } |
There was a problem hiding this comment.
This isn't needed since we'll just skip the for loop when there are no exclusions
| .await; | ||
| }); | ||
| } | ||
| } |
There was a problem hiding this comment.
I find this a little odd to have the optional dev_ctx passed into run. Another approach would be to have a new function start that does most of what run does now except for the last line graph::futures03::future::pending::<()>().await so that graph-node's run function would call start(..) and then wait on that pending future whereas gnd's run would call start and then block on watch_subgraph_updates
| .with_initdb_param("-E", "UTF8") | ||
| .with_initdb_param("--locale", "C") | ||
| .start_async() | ||
| .await; |
There was a problem hiding this comment.
It would be a nice enhancement if it was possible to pass a postgres URL in as an option, so that when you already have PG setup you can just use that, and use pgtemp if the user doesn't pass a URL in.
Follow up for #5977