Closed zachary-foster closed 7 years ago
I agree that any one person will likely stick to one way or the other.
I think we should stick with what you've done for now - we should get others to take a look at the entire package soonish and if there's a strong feeling about changes, we can talk about that then
Sounds good
Do we still want to have others weigh in on this? I can ask a few coworkers.
yeah, that'd be great
Hey @knausb and @Tabima, would you mind weighing in on this if you have time? Thanks!
Oh yea, and we would of course like to hear from @grunwald as well.
I think its great that you have engineered both behaviours! I think that will make the package approachable to those of us who have a more more 'base R' perspective as well as those who want the side effects version.
Thanks for the feedback @knausb!
I agree with all comments and support the dual approach.
Think we can close this or do we want more input?
i think we can close
Cool
@sckott, currently, I have set up the two ways of calling functions to behave differently. Calling a function in the classical R way imitates traditional no-side-effects copy-on-change behavior by cloning the object before returning the changed clone version. For example,
filter_taxa(ex_taxmap, 1:3)
will not changeex_taxmap
, butex_taxmap$filter_taxa(1:3)
will. Both return the modifiedex_taxmap
.Do you like this convention, or do you think it will confuse users? I suspect that each user will pick the style of calling they like and stick to it for the most part.