Plugins may introduce some risks imo.
Non-standard behaviour may be a b*tch.
E.g. the idea of a plugin which posts tags:
How are these elected and shares across instances? And displayed on clients?
Are they modifying the actual data written by the user in order to sync?
Maybe they are attractive to admins.
But they can mostlikely already query and modify the database, right?
I do not want to be against it just mentioning that it may introduce problems on its own which in turn needs to get adressed.
E.g.: When multiple plugins do a task at the same hook; How is the ordering managed?
When are transactions committed?
Should there be a maximal amount of time spend on plugins at some hook? How are resources shared then?
Let’s think about bad actors:
Meta deploys provides a plugin which compresses and decompresses post content and saves plenty of ressources for the admin. After a couple of years they put it to the grave or change the compression methods such that old posts cant be retrieved. But their instance surely still can access those.
I admire beeing lean. Had some projects where bad plugins raised in popularity and become the defacto standard. But they were resource-hungry and badly written or barely maintained. Workarounds spread back to the original program.
Just looked the first time into the lemmy code and it appears to be very neat and clean. I would recommend to stick to it.
But then I am no maintainer and a nobody shrugs
//edit: To me plugins are good to aid customization and enlarging the user base. I do not see how this contributes to the fediverse and instances in the long run.
Plugins may introduce some risks imo. Non-standard behaviour may be a b*tch.
E.g. the idea of a plugin which posts tags:
How are these elected and shares across instances? And displayed on clients? Are they modifying the actual data written by the user in order to sync?
Maybe they are attractive to admins. But they can mostlikely already query and modify the database, right?
I do not want to be against it just mentioning that it may introduce problems on its own which in turn needs to get adressed. E.g.: When multiple plugins do a task at the same hook; How is the ordering managed? When are transactions committed? Should there be a maximal amount of time spend on plugins at some hook? How are resources shared then?
Let’s think about bad actors: Meta deploys provides a plugin which compresses and decompresses post content and saves plenty of ressources for the admin. After a couple of years they put it to the grave or change the compression methods such that old posts cant be retrieved. But their instance surely still can access those.
I admire beeing lean. Had some projects where bad plugins raised in popularity and become the defacto standard. But they were resource-hungry and badly written or barely maintained. Workarounds spread back to the original program.
Just looked the first time into the lemmy code and it appears to be very neat and clean. I would recommend to stick to it. But then I am no maintainer and a nobody shrugs
//edit: To me plugins are good to aid customization and enlarging the user base. I do not see how this contributes to the fediverse and instances in the long run.