Where communities thrive


  • Join over 1.5M+ people
  • Join over 100K+ communities
  • Free without limits
  • Create your own community
People
Repo info
Activity
  • 16:24
    codecov[bot] commented #809
  • 16:16
    scala-steward opened #809
  • May 08 16:31
    harpocrates commented #794
  • May 08 13:06
    agboom commented #794
  • May 08 13:04
    agboom commented #794
  • May 05 20:33
    codecov[bot] commented #799
  • May 05 20:32
    codecov[bot] commented #790
  • May 05 20:32
    codecov[bot] commented #791
  • May 05 20:32
    codecov[bot] commented #792
  • May 05 20:24
    codecov[bot] commented #799
  • May 05 20:24
    codecov[bot] commented #792
  • May 05 20:24
    codecov[bot] commented #791
  • May 05 20:24
    scala-steward synchronize #792
  • May 05 20:24
    codecov[bot] commented #790
  • May 05 20:24
    scala-steward synchronize #791
  • May 05 20:24
    scala-steward synchronize #790
  • May 05 20:24
    scala-steward synchronize #799
  • May 05 19:33

    julienrf on master

    Update sbt-scoverage to 1.7.3 (compare)

  • May 05 19:33
    julienrf closed #806
  • May 05 19:32

    julienrf on master

    Update jwt-play to 7.1.4 (compare)

Adriaan Groenenboom
@agboom
Also: I think the (optional) trialing slash I'm trying to implement will involve a change in Url.directive in the server interpreter, because Directive.Segments does not match trailing slashes: https://github.com/endpoints4s/endpoints4s/blob/154e7ce630f284056f8c9ba3d8f302e326c9d696/akka-http/server/src/main/scala/endpoints4s/akkahttp/server/Urls.scala#L64
The Scaladoc string of Directives.Segments says: "If the path has a trailing slash this slash will not be matched."
2 replies
Adriaan Groenenboom
@agboom
If I run sbt test on the latest commit, some tests appear to fail (only due to exceptions, not assertions). Is that expected behavior? Here are the logs: https://pastebin.com/mM7ykeav
In short, there's a timezone error (probably because of the tz config on my local machine?), a ServerListenException on port 9000 (I checked if the port was in use, but it wasn't) and a generic Exception with "Something went wrong" followed by a dead letter warning (but that appears to be part of testing the exception flow, so that's expected)
Julien Richard-Foy
@julienrf
@agboom, the JS tests need jsdom to be installed locally ($ npm install jsdom). The ServerListenException is not expected, and I think we also get it spuriously on the CI, so this is something we need to fix, and the “Something went wrong” is not an actual test failure, it’s part of the expected test execution.
1 reply
Adriaan Groenenboom
@agboom

@julienrf I think I found a fairly simple solution for the trailing slash feature. I've created a draft PR to demonstrate it: endpoints4s/endpoints4s#742
Please let me know if this is the right direction (some tests will still fail because it involves changing the interpreters for the Urls algebra). The idea is that you can simply add a slash to a path like this:

endpoint(
  method = Get,
  url = path / segment[String]() / "",
  response = textResponse,
)

I like this because it's intuitive and idiomatic to endpoints' URL DSL. What do you think?

3 replies
I'd also like to explore ways to create endpoints that accept both a trailing slash and no trailing slash (an optional /). In REST endpoints it's not uncommon to accept both trailing and no trailing slash endpoints and let one redirect to the other. Do you think it's sensible to somehow build this into endpoints4s for ease of use?
Adriaan Groenenboom
@agboom
This message was deleted
Adriaan Groenenboom
@agboom
While adding to the EndpointsTestApi I noticed that some endpoints are only used in the server and some in the client test suites (e.g. deleteEndpoint only in server.EndpointsTestSuite and optionalEndpoint only in client.EndpointsTestSuite. I actually expected that all endpoints would be used in all tests, or might I be missing something?
2 replies
Adriaan Groenenboom
@agboom
Hi, I was looking at BuiltInErrors for a possible extension of the algebra and noticed that while the algebra is content type independent, the various interpreters always implement its methods with the application/json content type. My thought then was: wouldn't it be better to use jsonResponse from JsonEntities for this to prevent repeating this pattern? It occurred to me that the reason might be because you then had to mix in JsonEntities in BuiltInErrors and the latter is used in all the Endpoints interpreters it would enforce a possibly unwanted dependency on the user. Is that correct or is there some other reason?
3 replies
Adriaan Groenenboom
@agboom
Sorry for asking all these questions, but I have another one (please tell me if I need to stop :see_no_evil:). Did you at one time consider to add support for redirect endpoints? Currently, I need to explicitly add any redirect url to the server implementation. This feels somewhat awkward and of course the documentation or client support is left out. On the other hand, I can imagine that it's kind of a niche use case. When would you want to document redirects with OpenAPI, or support it in a client? In my case I have a server that is both a JSON API and an HTML staticfile server, of which the latter has redirects. For simplicity and clarity I'd like to define both with endpoints4s. I guess that's kind of a use case, be it a niche one. Another use case I can think of is API endpoints redirecting trailing slashes to no trailing slashes (e.g. /api/v1/users/ to /api/v1/users) or vice-versa. This is often done with APIs to make the endpoints more discoverable (not fail on a single slash). Both use cases can be implemented in pure server code of course, but I think it would be cool to use the clean interface of endpoints4s for this if possible. What do you think?
22 replies
Dominic Egger
@GrafBlutwurst
I noticed something recently when looking at our logs. Is it just me using it wrong or does handleClientErrors suffer from the fact that you don't know what route actually failed. e.g. in POST /foo/bar/ the body failed to decode, you won't know it was on POST /foo/bar
Julien Richard-Foy
@julienrf
@GrafBlutwurst Which server do you use?
Dominic Egger
@GrafBlutwurst
http4s
but isn't it an issue of the algebra because handleClientErrors only gets an instance of Invalid ?
Julien Richard-Foy
@julienrf
This method is not defined in the algebra, only in the http4s server interpreter
But I agree that it seems impossible to know the route that failed from there
(ftr, the equivalent method in the akka-http interpreter does provide a way to retrieve the underlying route)
So, in the case of the http4s server, I agree that we should change the signature of handleClientErrors to receive more information about the underlying request
Dominic Egger
@GrafBlutwurst
ahh yeah i see it's on the http4s package. what would you thinkg the signature should be? then I can maybe do an MR soon :tm: (whenever the ticket for the logging enhancement lands in my sprint)
Julien Richard-Foy
@julienrf
I guess we want to add a Request parameter to the method handleClientErrors?
Dominic Egger
@GrafBlutwurst
hmm I suppose that's the way to go yeah. that makes sense. I'll have a sprint planning today and will lobby for me to grab that ticket so I can do an MR soon
Julien Richard-Foy
@julienrf
Thanks!
Dominic Egger
@GrafBlutwurst
fun fact. -XX:+CMSClassUnloadingEnabled in .jvmopts breaks sbt 1.4.7 with my java version since it seems to be an unknown argument
that being Java 15.0.1 do we need that flag?
Julien Richard-Foy
@julienrf
To be honest I don’t remember why it was added. Just try without it.
Dominic Egger
@GrafBlutwurst
endpoints4s/endpoints4s#755 done. yeah i threw out the VM arg. I think it's a common flag for SBT speedup but it works just fine without it.
Dominic Egger
@GrafBlutwurst
I killed bincompat with the signature change though. sorry :|
Julien Richard-Foy
@julienrf
It’s OK, we can bump the major version of the http4s interpreter, unless this is a problem?
Dominic Egger
@GrafBlutwurst
not for me certainly. I adjusted the MR accordingly
Dominic Egger
@GrafBlutwurst
re #755 should I do the same for serverErrors? or would you prefer a seperate MR?
3 replies
Dominic Egger
@GrafBlutwurst
Is there a good way to add the capability of tapping and transforming on an algbera? e.g. I want to log requests and responses, or add authentication? sure i could do that on http4s directly (except that I don't have control over the log format there).

The way i did that right now is by having an intermediate trait with

  protected def authenticatedRequest[UrlP, BodyP, Out](
    method: Method,
    url: Url[UrlP],
    entity: RequestEntity[BodyP]
  )(implicit
    tuplerUB: Tupler.Aux[UrlP, BodyP, Out]
  ): Request[ApiEndpoints.AuthenticatedMessage[Out]]

and that works fine for requests. but one of the issues of response logging is that the process of encoding a response is not effectful

Julien Richard-Foy
@julienrf
@GrafBlutwurst Most of the types support the xmap operation for transforming them. However, xmap takes two pure functions only, not effectful functions.
Dominic Egger
@GrafBlutwurst
Yeah that's what makes it tricky. some functionallity that we currently treat as pure would become effectful
I am even considering operating on the algebra itself and changing the way the implementedByEffect works to accomodate this
Julien Richard-Foy
@julienrf
So, I think the only way to go is to do like you did. Define your own methods that wrap the existing algebra, and that do their side-effects. But this might need some changes in how the http4s interpreter is defined, since not all types can be interpreter with effectful computations, as you noticed…
Dominic Egger
@GrafBlutwurst
but it'd be very cumbersome I think
maybe a new abstract type member that captures the effect on encoding?
then what I want would be much easier, overwrite that, add effectful logging and i should be done. might need some annoying machinery though to make this work though
Adriaan Groenenboom
@agboom
Don't know if this is related, but are there ways in the current algebra to compose different endpoints? For example, we have multiple endpoints that add functionality to an endpoint, such as authenticatedEndpoint[A, B]: Endpoint[A, Either[AuthError], B], endpointWithError[A, B]: Endpoint[A, Either[ClientError, B]] and versionedEndpoint[A, B](version: ApiVersion): Endpoint[A, B]. Because the Endpoint type is abstract, I cannot find a way to compose these like (pseudo-code): authenticateEndpoint(endpointWithError(versionedEndpoint(...))).
One way we thought of is to instantiate and Endpoint to a class like Endpoints(request: A, response: B, url: String,...) and then "rewrite" these in the interpreters to the corresponding instances (e.g. Route for Akka HTTP server or A => Future[B] for client). But I have a feeling this would involve a pretty fundamental change to the algebras so I'm certainly open to other ideas.
23 replies
This message was deleted
Adriaan Groenenboom
@agboom
What is exactly the use of def uri in the Akka HTTP interpreter? It doesn't seem to be used in any code, except via-via in a single line in one test. Could it be possible that the function is not needed? https://github.com/endpoints4s/endpoints4s/blob/master/akka-http/server/src/main/scala/endpoints4s/akkahttp/server/Endpoints.scala#L50
5 replies
Domantas Petrauskas
@ptrdom
Hi, I'm trying to figure out if Scala.js frontend + Akka-HTTP backend could use chunked JSON entities for both requests and responses. Seems like chunked JSON entities aren't implemented under endpoints.xhr, so it is not supported, right?
Julien Richard-Foy
@julienrf
@ptrdom Hello, you’re right ChunkedEntities is not supported by endpoints.xhr. I don’t remember why I didn’t include it, to be honest. I also remember that I created #287 when I was working on it, so maybe I faced some limitations when trying to implement ChunkedEntities in the xhr interpreter.
Domantas Petrauskas
@ptrdom
@julienrf Thank you for the reply. I am really interested in having a capability of bidirectional streaming in API. grpc-web should eventually support this, but there is no clear timeline for client-side streaming support, and protobuf model does seem unnecessarily limiting for traditional web API, compared to JSON. It would be an interesting endeavor to implement endpoints4s/endpoints4s#287, maybe I could give it a shot.
Julien Richard-Foy
@julienrf
@ptrdom That would be a great contribution, although it might require some significant work! By the way, I’ve just found this old attempt at implementing an xhr interpreter for streaming: https://github.com/endpoints4s/endpoints4s/compare/streaming-support#diff-fe8014ca568bf3fc852541330ae13649cf8d4e298abea4ce777f73ddeabd33e2.
I don’t fully remember if eventually I didn’t include it in my final work on streaming because it was not working, or because I didn’t have the resources to finalize the implementations
See also the PR that introduced streaming support, where I give some implementation details: endpoints4s/endpoints4s#465
Domantas Petrauskas
@ptrdom
@julienrf Neat, I can also see that you attempted implementing WebSocket endpoints. I'll check these references out before I start, hoping this will result in an actual contribution.