Embed this noticeilja (ilja@ilja.space)'s status on Monday, 12-Sep-2022 20:11:50 JST
iljaLast week I started working a it more ttd on admin-fe, but I'm honestly wondering how much sense it makes for what I'm doing. I'm basically writing a ton more code and spending time on the test than on the actual functionality. And the functionality isn't even that much. It's just hiding certain elements in specific cases, that's it...
@ilja it's a struggle because admin-fe basically has no tests and is a mess; if you want, you don't have to implement it, just document what to do and i'll try doing that!
@ilja writing the feature; don't expect to have proper tests in there considering the state of admin-fe
it's the software that gives timeout errors when downloading emoji packs after all, and the same that needs a cache flush when enabling db-based config for menus to show :cirnoHeh:
@helene I think I'll just drop the tests for the other things that still need to happen. At least the "hide this element" ones. I also don't think these changes really improve admin-fe at all :/ They're needed for the changes in BE, but the whole thing just doesn't seem like it was really made for it and I'm just hacking it to make it work...
When a logged-in user isn't privileged, they shouldn't see whatever buttons or drop-downs allow them actions they aren't privileged too. The issue lists what still needs to be done for those. Then there's also the fact a bunch of calls are being made, but a lot can return a 403, so we need a way to handle that.
The rest of the hiding, I can do since I'm already working on it. The calls is something I still have to figure out.
@helene Not sure if it has to be BE, but yeah, I had the same feeling when first poking at it :/ There's probably stuff in there that can be reused, but there's stuff in there that don't even look like it's being used. It also seems like it was written to go in a certain direction, but that direction is different of where we're going, so it's like going against the tide with that...
@ilja to me, having it in BE would be best especially if we want to get rid of FE bundles, and LiveView does allow us to basically not have to deal with the JavaScript ecosystem by writing everything in Elixir and on the server, which I feel is perfect when wanting to easily extend things and make good admin tools as the features grow It doesn't matter for it to be client side, it's all a panel for controlling the backend, so LiveView really would make sense
LiveView seems interesting ? We probably still want things to be accessible via API (idk why, but somehow I think that's important), but having a build in FE doesn't exclude that. Having something that still allows to easily extend things seems good to me, and it would also allow us to fix the issue of having way too technical language and structure in the admin parts.
@ilja@helene Literally sold on doing this. I experimented with doing that for my Fuck Gab! project. And it actually is pretty promising as a technology.