Find here answers to the most common questions that we get asked about http4k
Q. Is http4k a library or a framework?
A. Although it has many of the features of a framework, we consider http4k to be a library which adds a common HTTP routing layer. It is incredibly unopinionated and has been designed to not enforce design decisions on the API user. We use http4k for applications both large and small, using no DI framework.
Q. Is http4k currently used in production?
A. Absolutely! Whilst overall stats are obviously hard to come by, the biggest known usage of the library is serving the global site traffic (rank ~700 globally) for a large academic publisher, easily serving 10s of millions of requests per day on a few nodes. Additionally judging from the download stats and interest in the Slack channel indicate that take-up is increasing nicely. http4k also appears in the Thoughtworks Tech Radar, which covers useful and upcoming technologies used in Thoughtworks-delivered projects.
If you're running http4k in production and would like to be listed on the site as an adopter, please get in touch.
Q. Does http4k support an Async model? I need webscale!
A. Not at the moment. Adding Async support is a decision that we are thinking about carefully so that we don't end up complicating the API. When we do add it, it'll use co-routines model so we also want to ensure that the integrations with the various backends and clients are solid, as well as supporting essential features that currently rely on Threads, such as Zipkin Request Tracing and Resilience4j support. As for the scaling arguments, see the above answer relating to production usage, or checkout the benchmark results to see how http4k compares to other JVM-based sync and async web libraries.
Q. I'm attempting to build HTTP messages using the API, but changes don't affect the object (e.g. calling
A. http4k HTTP message objects are immutable, so you need to chain or reassign the value from the method call to get the updated version.
Q. Where are all the useful Filters defined?
A. Filters are all in the
import org.http4k.filter package and are located as methods on a singleton
object relevant to their use:
Lenses & Auto-Marshalling
Q. Where is the
Body.auto method defined?
Body.auto is an extension method which is declared on the parent singleton
object for each of the message libraries that supports auto-marshalling - eg.
Xml. All of these objects are declared in the same package, so you need to add an import similar to:
Q. Declared with
Body.auto<List<XXX>>().toLens(), my auto-marshalled List doesn't extract properly!
A. This occurs in Jackson and Moshi when serialising bare lists to/from JSON and is to do with the underlying library being lazy in deserialising objects (using LinkedHashTreeMap) ()). Use
Body.auto<Array<MyIntWrapper>>().toLens() instead. Yes, it's annoying but we haven't found a way to turn if off.
Q. Using Jackson, the Data class auto-marshalling is not working correctly when my JSON fields start with capital letters
A. Because of the way in which the Jackson library works, uppercase field names are NOT supported. Either switch out to use
http4k-format-gson (which has the same API), or annotate your Data class with
@JsonNaming(PropertyNamingStrategy.UpperCamelCaseStrategy.class) or the fields with
@JsonAlias or to get it work correctly.
Q. Using Gson, the data class auto-marshalling does not fail when a null is populated in a Kotlin non-nullable field
A. This happens because http4k uses straight GSON demarshalling, of JVM objects with no-Kotlin library in the mix. The nullability generally gets checked at compile-type and the lack of a Kotlin sanity check library exposes this flaw. No current fix - apart from to use the Jackson demarshalling instead!
Q. My application uses Lenses, but when they fail I get an HTTP 500 instead of the promised 400.
A. You forgot to add the
ServerFilters.CatchLensFailure filter to your Server stack.