Simon Hengel

Corresponding code: https://github.com/haskell/text/blob/5e57460711a9a5ab7f8a30f0e11cd850018dae70/src/Data/Text/Internal/Search.hs#L49-L56 https://github.com/haskell/text/blob/5e57460711a9a5ab7f8a30f0e11cd850018dae70/src/Data/Text/Internal/Search.hs#L93-L100

> When the response-data has already been read into the buffer, the connection is closed Are you saying that after 1. `connectionRead` 1. `connectionUnread` 1. `connectionClose` you expect a subsequent...

> but since the connection has already been closed, an exception is thrown and the body remains unread Do you know where the connection is closed?

@adithyaov sorry, I'm still a little bit lost here. Am I right to assume that you consider this a bug fix (as opposed to a new feature)? As for [cachix/hs-opentelemetry-instrumentation-http-client](https://github.com/cachix/hs-opentelemetry-instrumentation-http-client),...

> > But every bug report and every feature request starts from the user's perspective, and I'm missing the user's perspective here. > > I don't understand what you mean...

> ## Side effects of this change > `mkManagerSettingsContext'` was originally idempotent; this change breaks that idempotency. > > The only issue I see is that multiple applications of `mkManagerSettingsContext'`...