| 15 Mar 2023 |
 hexa | yep 🙂 | 12:00:36 |
| hexa | make the default closure small | 12:00:44 |
| hexa | make sure extras and plugins don't cause a rebuild, so everyone can rely on the cached package | 12:01:01 |
| hexa | Redacted or Malformed Event | 12:01:54 |
| hexa | Redacted or Malformed Event | 12:02:03 |
| hexa | prometheus_client is actually not an optional dependency | 12:06:47 |
| hexa | the idea is to make it more feasible to expose these extras listed here https://github.com/matrix-org/synapse/blob/develop/pyproject.toml#L249-L273 | 12:07:17 |
| hexa | which are all the packages marked as optional in the dependencies section | 12:07:27 |
| hexa | but grouping them like this makes their use much clearer | 12:07:37 |
| hexa | and allows us to select them intelligently | 12:07:44 |
 ma27 | In reply to @hexa:lossy.network
prometheus_client is actually not an optional dependency heh, that was a wild guess and served just as an example :) | 12:10:14 |
| hexa | yeah, I feel the same about prometheus, why isn't it optional? | 12:10:34 |
| ma27 | but yeah it sounds reasonable, so I'd say if you have the time for it then go for it, I'm happy to test the changes on my server if needed then :) | 12:10:55 |
| hexa | fine, lets get the bump in first, I'll rebase 😛 | 12:38:11 |
| hexa | https://github.com/NixOS/nixpkgs/pull/221318 | 12:52:20 |
| hexa | Dandellion: by the way: worker support when? 😄 | 12:59:27 |
 Dandellion | 😬 probably after I've reworked how I do stream writers | 13:02:17 |
| hexa | stream writers? | 13:21:04 |
| Dandellion | https://github.com/matrix-org/synapse/blob/develop/docs/workers.md#stream-writers
they're just a little more involved to set up so aren't as compatible with some of the abstracted worker-type machinery | 13:25:57 |
| hexa | hrm | 13:27:23 |
| hexa | I would be up for a smaller start and incrementally adding complexity | 13:27:37 |
| hexa | because so far we only have out-of-tree solution | 13:27:48 |
| hexa | * because so far we only have the out-of-tree solution | 13:28:14 |
| Dandellion | I can just upstream the worker.instance stuff first ig | 13:28:15 |
| hexa | probably for the best | 13:28:25 |
| hexa | would be happy to review | 13:28:49 |
| Dandellion | maybe this weekend then :) | 13:28:53 |
| 19 Mar 2023 |
 motiejus | hi folks. I am setting up Synapse on my brand new NixOS server (still consider myself new to it). I have a couple of secrets -- registration_shared_secret, turn_shared_secret, macaroon_secret_key, which I am supposed to pass over extraConfigFiles`. Here is how I envision the process exactly:
- The secrets are written on ramfs reachable as root alone.
ExecStartPre reads all secret files and pre-populates the yamls somewhere in /run/synapse via dirty printfs.extraConfigFiles then point to the places wherever the one-step-above generates them.
Am I doing this right? Any other hints/ideas?
| 19:11:28 |
| motiejus | * hi folks. I am setting up Synapse on my brand new NixOS server (still consider myself new to NixOS, but have been maintaining my synapse for ~2 years). I have a couple of secrets -- registration_shared_secret, turn_shared_secret, macaroon_secret_key, which I am supposed to pass over extraConfigFiles`. Here is how I envision the process exactly:
- The secrets are written on ramfs reachable as root alone.
ExecStartPre reads all secret files and pre-populates the yamls somewhere in /run/synapse via dirty printfs.extraConfigFiles then point to the places wherever the one-step-above generates them.
Am I doing this right? Any other hints/ideas?
| 19:11:48 |
| motiejus | * hi folks. I am setting up Synapse on my brand new NixOS server (still consider myself new to NixOS, but have been maintaining my synapse for ~2 years). I have a couple of secrets -- registration_shared_secret, turn_shared_secret, macaroon_secret_key, which I am supposed to pass over extraConfigFiles\. Here is how I envision the process exactly:
- The secrets are written on ramfs reachable as root alone.
ExecStartPre reads all secret files and pre-populates the yamls somewhere in /run/synapse via dirty printfs.extraConfigFiles then point to the places wherever the one-step-above generates them.
Am I doing this right? Any other hints/ideas?
| 19:12:03 |
