possible performance problems

[alosdev]

https://github.com/rybalkinsd/kohttp/blob/188b5c5a6e770a167fabcddc928dd4978af2a230/kohttp-jackson/src/main/kotlin/io/github/rybalkinsd/kohttp/jackson/ext/ResponseExt.kt#L69

in the following line, the library makes a string out of the response and it loses the purpose of a stream reader. Would it be possible to use the InpputStream to leverage the way how Jackson is working? Also with very large responses, it would raise some memory pressure.

[rybalkinsd]

Thanks for posting this issue @alosdev!

Overall, It's a great idea to introduce a simplified API to work with response streaming.

Indeed, in the quoted file we load the whole response. The main idea of .toType extension is to deserialise simple/small bodies as type T

val simple: SimpleClass? = it.toType()

In this case we have an extra allocation(for tmp string) + sequential processing vs streaming processing overhead. In my mind it would be great to have a benchmark to make a decision. Do you have any use case in your mind?

From my understanding getting deserialized type T or JsonNode itself will not radically change the situation (the whole object will be in memory after parsing) However parsing arrays of to Stream<T> or fetching a particular filed(sub-object) from a huge body could help. Do you have any thoughts about API?