You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
This stems from a concern that the provided interfaces prevent reuse of what should be immutable objects in memory. For context, to load up a SchemaRepository's lookup table, we need to read in our schema files and then run them through a dereference(). The OpenAPIContract class does that alongside some validation in its from() function, which is really nice.
To explain the problem, now suppose that I have multiple open api files and thus need multiple OpenAPIContracts that produce routers that are eventually attached as subrouters for the rest server. Each OpenAPIContract will have its own SchemaRepository. Also consider that someone may want a fully hydrated SchemaRepository that I want to expose to a swagger ui or arbitrary http clients (I did that with a handler on some schema route that does the lookup/resolve for the schema and returns it). Every single OpenAPIContract and the separate SchemaRepository for exposing to clients will have duplicate/immutable schemas in memory.
The feature would be to genericize the validation/repo filling (which is filled by calling dereference() on the repo) logic in OpenAPIContract::from, probably moved into the vertx-json-schema library. Then allow a SchemaRepository to be provided to OpenAPIContract::from. That would allow the hydration of a SchemaRepository and then for it to be reused everywhere it is needed.
Use cases
The big use case is the reuse of SchemaRepository's for OpenAPIContracts and also the generalization of a very useful hydration feature for loading schemas into a SchemaRepository. OpenAPIContract is not the only class that can make use of that validation/repo filling logic. The only useful logic I've seen for filling a SchemaRepository's internal lookup map is in OpenAPIContract, but that's not something that should be specific to the openapi library.
Code reuse and saving on memory usage and startup time are the real benefits. In my eyes I would say that this is necessary polish.
Contribution
Who should implement this feature ? are you volunteering for implementing this feature or
do you know that is able and willing implement this feature ?
The text was updated successfully, but these errors were encountered:
Hi @jawaff thanks for bringing this up. Did I understand it correct, that you would like to create one SchemaRepository and reuse it in multiple instances of OpenAPIContract?
Maybe you can instantiate OpenAPIContractImpl by yourself and simply pass your SchemaRepository. Would this solve your problem? Of course this isn't a public API, but at the moment I don't see anything that would make this constructor deprecated.
Describe the feature
This stems from a concern that the provided interfaces prevent reuse of what should be immutable objects in memory. For context, to load up a
SchemaRepository
's lookup table, we need to read in our schema files and then run them through adereference()
. TheOpenAPIContract
class does that alongside some validation in itsfrom()
function, which is really nice.To explain the problem, now suppose that I have multiple open api files and thus need multiple OpenAPIContracts that produce routers that are eventually attached as subrouters for the rest server. Each OpenAPIContract will have its own SchemaRepository. Also consider that someone may want a fully hydrated
SchemaRepository
that I want to expose to a swagger ui or arbitrary http clients (I did that with a handler on some schema route that does the lookup/resolve for the schema and returns it). Every single OpenAPIContract and the separate SchemaRepository for exposing to clients will have duplicate/immutable schemas in memory.The feature would be to genericize the validation/repo filling (which is filled by calling dereference() on the repo) logic in
OpenAPIContract::from
, probably moved into the vertx-json-schema library. Then allow a SchemaRepository to be provided toOpenAPIContract::from
. That would allow the hydration of a SchemaRepository and then for it to be reused everywhere it is needed.Use cases
The big use case is the reuse of SchemaRepository's for OpenAPIContracts and also the generalization of a very useful hydration feature for loading schemas into a SchemaRepository. OpenAPIContract is not the only class that can make use of that validation/repo filling logic. The only useful logic I've seen for filling a SchemaRepository's internal lookup map is in OpenAPIContract, but that's not something that should be specific to the openapi library.
Code reuse and saving on memory usage and startup time are the real benefits. In my eyes I would say that this is necessary polish.
Contribution
Who should implement this feature ? are you volunteering for implementing this feature or
do you know that is able and willing implement this feature ?
The text was updated successfully, but these errors were encountered: