Why source plugins try to infer data instead of feching the structure from CMS backend

[lucashfreitas]

I have seen many people facing the same issue and have personally faced it many times in the past.

I am not 100% sure why the source plugins uses this "automatic" way of inferring data types if the CMSs always have the data structure in their end and it's the source of truth. Why infert types based on the data returned instead of reading the data structure from CMS and building the graphql schema? I know we can customize schema do get around this but would be a bit overkilling to do that manually for every content type, specially for big websites. I have been using gatsby-source-graphql in favor of other source plugins because it's awesome and have really good integration with typescript and can auto generate types using graphql-code-gen plugins, etc.

Let me describe the problem:

let's supose we have 10 entries that has a title property but the user didn't fill any of them, if you try to query this field using the the majority of source-plugins (including gatsby-source-contentful), it will return an error saying that the field doesn't exists. Considering that the structure,field types can easily be retrieved from the CMS and it is the single source of truth I am not 100% sure why the plugins try to infer graphql schema types instead of building the nodes based on the content type according to each CMS.

[lucashfreitas]

After chatting about that with @KyleAMathews the answer is that some plugins like Sanity, DatoCMS, Wordpress already do that. Contentful is planning to do that in the major release #31385