ZPipeline.fromFunction introduces a significant performance overhead about zio HOT 6 CLOSED

They're not currently, just ran on a manual basis.

from zio.

Contribution would be welcome 👍🙏

from zio.

Contribution would be welcome 👍🙏

👍 do u happen to know if there's a way to run the benchmarks as part of the PR's CI?

from zio.

seems like ZChannel.Bridge is now only used by the ZChannel.mergeXXX family of functions, not really justified I might add.
these functions allow all merged channels to read from the same upstream, something similar to broadcast followed by merge, in fact most if not all stream operations involving merge operate on streams (one of the reason for introducing ZPipeline.fromFunction) so user's would have to work pretty hard to come up with code that utilizes the 'shared' upstream.

it's worth noting that the channel executor identifies this and skips the bridge in these scenarios.

so two questions here:

1. do we want to introduce the mergeXXX functions on ZPipeline as well?
2. can we completely deprecate the Bridge (I recall Adam stating on Discord he'd like to get rid of it), if so how do we replace and deprecate the ZChannel.mergeXXX family of functions which are currently public API

from zio.

Closed by #8761

from zio.

Since Bridge is private to zio, I think we could potentially delete if we replace its current usage by something else, but all public API need to remain.

from zio.