Fork me on GitHub
#pedestal
<
2020-06-11
>
orestis05:06:30

I don’t use async handlers today so I guess my code will look the same :)

orestis12:06:50

For the life of mine I can’t figure out why upgrading lacinia-pedestal to 0.13.0 breaks GraphiQL for me. I’ve narrowed it down to com.walmartlabs.lacinia.pedestal/graphiql-ide-response which somehow generates connectionParams: --NO-MATCH--which of course the browser complains about — but even if I pass in ide-connection-params as an option I get the same result.

orestis12:06:05

I’ve literally copy-pasted the function definition and when I call it locally, everything works as expected.

orestis12:06:57

Huh. Of course the moment I try to attach a debugger everything works 😞

orestis12:06:12

But not from a fresh VM. I’m not sure what’s going on.

orestis13:06:56

Ah, wrong channel, was meant to go to #graphql — sorry

hindol14:06:56

@orestis git bisect really helps in such cases. For a deps.edn based library and if you yourself use deps, you can point to a local copy of the library and run git bisect. But here I see they are using lein. I wonder if there is still a way to bisect.

orestis14:06:49

Turns out that (as expected) it was a problem on my end (AOT compiled deps)

orestis14:06:57

Having fixed my AOT problems and now actually using pedestal.log 0.5.8 I’ve found a breaking change which should be impacting everyone: https://github.com/pedestal/pedestal/issues/638#issuecomment-642714551

orestis14:06:02

Seems like it’s only direct use of log/log that breaks it; log/info still works.

hindol15:06:23

Good thing you figured it out. These types of bugs are always difficult.

ddeaguiar16:06:02

@orestis sorry you ran into this issue. Creating an issue and will tackle tomorrow.

orestis16:06:54

No worries, I figured it out and have a workaround :)