marcandre
Is there a plan to support custom loading for embedded types?
Custom types do not use load/dump when embedded.
Then may use dump if they define embed_as to return :dump, but there is no way to use load or any other mechanism then cast when loading data.
- Is there a good reason for this design? The asymmetry in dumping / loading is a bit odd to me.
Moreover the fact that cast has to be called when loading from the DB can be quite inconvenient. While returning :error when casting user data (e.g. from a form) is very useful, if I have invalid data stored in my DB then returning :error result in a raised exception, which is blocking. Moreover, there’s no way to know within cast if I’m loading from the DB or casting user data.
- Documentation for
loadis very misleading, e.g.
There are two situations where this callback is called:
- When casting values by Ecto.Changeset
- When passing arguments to Ecto.Query
Clearly, this is incomplete and there is an extra situation “3. When loading data in an embedded schema”.
I can propose a PR but would appreciate to know if there’s every a plan to support custom loading for embedded types.
Thank you
Trending in Questions
Other Trending Topics
Categories:
Sub Categories:
Forums
Popular Tags
- #ecto
- #liveview
- #troubleshooting
- #learning-elixir
- #deployment
- #library
- #erlang
- #testing
- #genserver
- #mix
- #absinthe
- #remote-other
- #otp
- #plug
- #how-to-question
- #macros
- #postgres
- #channels
- #elixirconf
- #exunit
- #discussion
- #code-sync
- #javascript
- #podcasts
- #onsite
- #dialyzer
- #docker
- #authentication
- #umbrella
- #full-time-contract
- #podcasts-by-brainlid
- #ecto-query
- #elixir-ls
- #phoenix_html
- #iex
- #blog-post
- #graphql
- #genstage
- #ai
- #websockets
- #elixirconf-us
- #supervisor
- #advent-of-code
- #distillery
- #processes
- #api
- #forms
- #metaprogramming
- #performance
- #security










First 3 of 3 Posts!
arcanemachine
Maybe try posting this on the Ecto dev group as well.
josevalim
We can always add a new value to em embed_as, we just need to understand the use case. The reason it was designed like this is because most embedding is done as JSON, which is a lossy format. And PRs to improve the docs are always welcome.
marcandre
Thank you. I went another way for now, so I’m not 100% confident about my usecase
. I created a PR for the doc, thanks 