You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
It is a bit of a smell to have unused timely dataflow streams. They are not harmful (data terminate at operators that are not used, rather than eagerly cloned for an unused stream), but it is odd to create one and not use it. We could mark them as MustUse which will at least warn if streams are created and then dropped without being used.
The main downside seems to be that it would introduce warning noise to all timely projects, as e.g. the pattern
mystream
.do_stuff()
.probe_with(&mut probe);
ends with a stream (probe_with outputs its input stream, after the probe has been updated). Perhaps we'd want a .cap() method to make it easy to remove the error. Putting a let _ = at the beginning of each of these pipelines might be annoying. Anyhow, food for thought.
It is a bit of a smell to have unused timely dataflow streams. They are not harmful (data terminate at operators that are not used, rather than eagerly cloned for an unused stream), but it is odd to create one and not use it. We could mark them as
MustUse
which will at least warn if streams are created and then dropped without being used.The main downside seems to be that it would introduce warning noise to all timely projects, as e.g. the pattern
ends with a stream (
probe_with
outputs its input stream, after the probe has been updated). Perhaps we'd want a.cap()
method to make it easy to remove the error. Putting alet _ =
at the beginning of each of these pipelines might be annoying. Anyhow, food for thought.cc: @benesch
The text was updated successfully, but these errors were encountered: