| 9 Jul 2020 |
 Timo | I meant if it's not common, the slight performance loss might be okay if the sending server doesn't have to keep track of so many things in RAM then | 10:49:20 |
| Timo | In reply to @richvdh:sw1v.org
to a certain extent, this is up to the implementation If that's the case, I'll experiment with optimizations | 11:29:09 |
|  ringo joined the room. | 20:31:03 |
| 10 Jul 2020 |
 richvdh | Kegan: so have you run complement against synapse yet? | 12:46:47 |
| richvdh | rav@fred:~/work/complement (master =)$ COMPLEMENT_BASE_IMAGE=complement-dendrite:latest go test -v ./tests
# github.com/matrix-org/complement/tests
tests/federation_keys_test.go:4:2: unknown import path "crypto/ed25519": cannot find module providing package crypto/ed25519
FAIL github.com/matrix-org/complement/tests [setup failed]
| 12:54:55 |
| richvdh | my go is rusty; any ideas? | 12:55:03 |
 tulir | what go version? | 12:56:30 |
| richvdh | rav@fred:~/work/complement (master =)$ go version
go version go1.11.5 linux/amd64
| 12:56:53 |
| richvdh | apparently | 12:56:55 |
| richvdh | that does sound a bit old doesn't it | 12:57:00 |
 neilalexander | You'll need 1.13 or later for crypto/ed25519 | 12:58:06 |
| tulir | yeah | 12:57:05 |
| neilalexander | * You'll need 1.13 or later for crypto/ed25519 | 12:58:08 |
| neilalexander | (our baseline across the board for dendrite, gmsl etc now is Go 1.13) | 12:58:29 |
| richvdh | ok thanks | 12:58:33 |
| richvdh | complement readme should probably say that 😇 | 12:58:51 |
| richvdh | now it seems to be failing to spin up docker containers | 13:26:58 |
| richvdh | is there a way to just run one test ? | 13:30:22 |
| richvdh | well, I get this:
rav@fred:~/work/complement (master =)$ COMPLEMENT_DEBUG=1 COMPLEMENT_BASE_IMAGE=complement-dendrite:latest go test -v ./tests
2020/07/10 14:33:09 config: &{BaseImageURI:complement-dendrite:latest BaseImageArgs:[] DebugLoggingEnabled:true}
=== RUN TestRegistration
2020/07/10 14:33:10 alice.hs1 : constructing homeserver...
2020/07/10 14:33:11 alice.hs1 : failed to deployBaseImage: alice.hs1: failed to check server is up. GET http://localhost:32775/_matrix/client/versions => error: Get "http://localhost:32775/_matrix/client/versions": read tcp 127.0.0.1:41022->127.0.0.1:32775: read: connection reset by peer
2020/07/10 14:33:11 ============================================
2020/07/10 14:33:11 alice.hs1 : Server logs:
2020/07/10 14:33:11 ============== alice.hs1 : END LOGS ==============
2020/07/10 14:33:11 : Failed to kill container : Error response from daemon: page not found
TestRegistration: apidoc_register_test.go:30: Deploy: Failed to construct blueprint: failed to find built images via ImageList: did they all build ok?
--- FAIL: TestRegistration (13.61s)
| 13:34:42 |
| neilalexander | Interesting | 13:36:04 |
| neilalexander | I've yet to try this in the real world so not sure if it's an isolated problem | 13:36:14 |
| richvdh | I think it's just not waiting long enough for the image to start | 13:39:04 |
| * richvdh adds a time.Sleep | 13:41:03 |
| richvdh | well now it's getting further | 13:42:28 |
| richvdh | I think I'm going to put this on hold until kegan can give me a hand | 13:43:20 |
|  maurice joined the room. | 19:33:12 |
| 11 Jul 2020 |
| Timo | https://github.com/matrix-org/sytest/issues/913 | 12:04:53 |
| Timo | https://github.com/matrix-org/sytest/issues/914 | 12:05:14 |
| Timo | Kegan: Here are some problems I found in sytest. They should be pretty easy to fix, but they currently hinder Conduit's sytest percentage a lot | 12:12:35 |
| Timo | https://github.com/matrix-org/sytest/issues/915 | 12:16:41 |
