| 23 Aug 2020 |
 jan | In reply to @andrewsh:matrix.org
the backports for 1.19 will take some time Well… that was fast! 👍️ | 17:10:16 |
 andrewsh | it was indeed a bit faster than I anticipated | 17:11:17 |
 grin | ERROR:root:Needed txredisapi>=1.4.7 for the 'redis' feature but it was not installed | 23:08:21 |
| grin | seems synapse is not happy about a non-debianised python lib | 23:09:09 |
 davo | In reply to @_neb_rssbot_=40hubert=3auhoreg.ca:matrix.org
Debian package news for matrix-synapse: Accepted matrix-synapse 1.19.0-1~bpo10+1 (source) into buster-backports ( https://tracker.debian.org/news/1170975/accepted-matrix-synapse-1190-1bpo101-source-into-buster-backports/ ) Wait what, it skipped migrating to testing? | 23:41:15 |
 emorrp1 | it's not a real copy from testing to backports, but a re-upload - usually backports ftp-masters don't accept a version that isn't in testing | 23:42:40 |
| 24 Aug 2020 |
| andrewsh | it didn’t skip anything | 08:55:39 |
| andrewsh | oh, in fact it did | 08:55:55 |
| andrewsh | In reply to @grin:grin.hu
ERROR:root:Needed txredisapi>=1.4.7 for the 'redis' feature but it was not installed I don’t see this error, what version of synapse is that? | 08:56:41 |
| grin | andrewsh: redis config is very hidden in the debian version, since it even includes an older example config | 09:04:38 |
| grin | while new vector/element/synapse insist that workers shall use redis | 09:05:02 |
| grin | so when you try to use redis: parts [homeserver.yaml] you get this. | 09:05:35 |
| andrewsh | well, it’s an optional thing, isn’t it? | 09:11:33 |
| andrewsh | (I should still package txredisapi, I guess) | 09:12:01 |
| 25 Aug 2020 |
| grin | As long as one
- doesn't use workers, and
- doesn't follow the very emphasized advice from Element, Inc. to use redis instead of "outdated tcp replication protocol"
I guess it could be said so.
| 11:46:03 |
| grin | * As long as one
- doesn't use workers, and
- doesn't follow the very emphasized advice from Element, Inc. to use redis instead of "outdated tcp replication protocol"
I guess it could be said so.
| 11:46:06 |
| emorrp1 | well yeah, I'd say it's optional given how long matrix homeservers survived without redis | 17:10:22 |
| emorrp1 | I wonder how debian applications implement recommend postgres, perhaps a matrix-synapse-postgres package which would configure pg users and drop in a config snippet? | 17:12:11 |
 jonas | most commonly, packages optionally using an SQL backend use dbconfig-common for that | 17:39:12 |
| jonas | ...and anyone then wanting to not rely on dbconfig-common to maintain their local SQL stuff can install package dbconfig-no-thanks | 17:40:36 |
| emorrp1 | jonas: thanks, that looks like exactly what I was after, maybe I can contribute a dbconfig-common integration for matrix-synapse | 17:41:34 |
| jonas | I would expect such patch proposal would be appreciated - but I am not working on that package, so if you are probing for pre-approval then I guess it is the opinion of andrewsh you should pay attention to | 17:44:22 |
| jonas | ...but really, I suggest to not wait for that but just go ahead and file a bugreport when you got a patch proposal - worst case you gained some experience with dbcommon that is useful elsewhere :-) | 17:45:20 |
| andrewsh | yeah, by all means | 17:45:35 |
| jonas | :-D | 17:45:43 |
| andrewsh | I *do* accept patches! | 17:45:46 |
| jonas | _any_ patches? ;-) | 17:45:57 |
| emorrp1 | BOBBY'; DROP TABLES; | 17:46:21 |
| andrewsh | dbconfig, redis, better config handling, initscripts — all of those are things we need to improve | 17:46:29 |
| andrewsh | jonas, I will certainly not accept patches making all installations belong to Dr Jonas ;) | 17:47:01 |
