currently it pre-indexes everything in the root. This behavior is undesirable out of the box, so it now looks recursively by default.
Signed-off-by: quobix <dave@quobix.com>
Document configuration has been simplified, no more need for AllowRemote stuff in the document configuration, it’s assumed by setting the baseURL or the basePath.
Signed-off-by: quobix <dave@quobix.com>
Every `Build()` method now requires a `context.Context`. This is so the rolodex knows where to resolve from when locating relative links. Without knowing where we are, there is no way to resolve anything. This new mechanism allows the model to recurse across as many files as required to locate references, without loosing track of where we are in the process.
Signed-off-by: quobix <dave@quobix.com>
There is a horrible amount of work to be done to clean this up, and wire in remote support. but so far, this is working as expected and is now a much cleaner design, (once everything has been cleaned up that is)
Signed-off-by: quobix <dave@quobix.com>
The index runs async everywhere, it's kinda impossible to know which path with resolve first, so testing is hard. Sometimes a race condition is hit, well, it was. Now that map has a mutex on it.
Also fully fixed handling files with relative links. A basepath property has been added to the index configuration to allow a local root to be set when resolving files.
Added a full checkout test for digital ocean so that full remote and full local testing is performed.
Thinking more about how to know what to resolve and what not to means depending on circular reference knowledge. So using an idea from @TristanSpeakEasy, the `*resolver.ResolvingError` now contains a pointer to the original circular error, and it also implements the `Error` interface correctly which allows it simple passage up the chain, without losing fidelity.
Added new documentation example as well
Signed-off-by: Dave Shanley <dave@quobix.com>
Tristan made a good point, part of the doc building process performs a resolving check and circular reference check, which is ignored by the returned errors. So resolving errors are now unpacked into standard errors and returned. Not sure why gofmt has shifted everything around however.
sending bolts of electricity through the entire stack is throwing errors here and there, simple null values that need checking, some forgotten has functions and what not. shaking out as many bad splices as possible as we reach for the summit.
The v3 model is wrong and out of sync with the spec. It's been corrected, so the v2 and v2 model for SecurityRequirement have been collapsed down into a base model., they are the same data structures. This has allowed me to delete the complexity of sharing two different models for the same structure, by unifying the model correctly. I am not sure why I decided to change the v3 model, oh well, its been corrected. Long live swagger!
3.0 and 2.0 do not work, there are multiple versions and anything with a period in it sucks from my point of view, v2 and v3 feel much better from a DX perspective.
