This fixes #121 but in an unsatisfying way. This seems to fix the flakiness of the selenium tests, but while inspecting the logs I noticed that the wasm/local/local-concurrent steps were actually all testing the wasm server. The servers compiled by cargo would exit when they noticed that port 8080 was busy, because the previous make serve-wasm was still running. Adding a kill %1 to ext.yaml, so that the servers are stopped in between steps, only made more issues crop up. I feel like this should be fixed but as a separate PR
This fixes #121 but in an unsatisfying way. This seems to fix the flakiness of the selenium tests, but while inspecting the logs I noticed that the wasm/local/local-concurrent steps were actually all testing the wasm server. The servers compiled by cargo would exit when they noticed that port 8080 was busy, because the previous
make serve-wasmwas still running. Adding akill %1toext.yaml, so that the servers are stopped in between steps, only made more issues crop up. I feel like this should be fixed but as a separate PR