Conversation
|
I'm not a fan of changing behaviour while doing a rewrite, now we'll have to figure out if ported tests are failing because of expected change, or because of some bugs introduced in the rewrite. I would like to postpone this kind of work (like for #21) until we migrated and released the first version of the package with a 1-1 behaviour with our current one. IMO we'll be better of:
|
|
I don't mind that we wait to merge then! 👍 |
I don't feel very strongly about it, but I wonder if there's merit to releasing all the breaking changes together. When we release this repo, it will already be coming with some breaking changes:
With this in mind, it might be better to also release other breaking changes like this one with it, so that the users can adjust once to the changes rather than several times. I don't mind which way we go, but wanted to add my 2c. edit: Merging after porting over tests makes sense regardless though! |
Implementation of a path connection strategy, described here: https://discord.com/channels/788353076129038346/914821651017908234/1217410551916003328
We now have a
connectPathwhich holds the models created before the current model, following the same "path".If in the current model, a parent is not specified, we look into the
connectPathfor a similar model name, if we find one we connect to it.The
connectoption takes precedence over the path strategy, so if you still want to use the path while connecting, it will have to be manual for now (using the connect callback).Fixes: S-2014