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
When writing data cleaning pipelines, I often find that single step requires more than 80 characters to do right, and to maintain readability, I split the expression over multiple lines. When I start a new pipeline expression, I would like to be able to have the indentation begin back at the original "left-most" point. For example, when I face this problem with dplyr in R, using emacs/ESS, I get indentation that looks like this:
However, the equivalent code in emacs/Julia-mode delivers indentation that looks like this:
Is this intended behavior? Is there a setting I can change to make Julia-mode indent this pipeline the way ESS indents the equivalent pipeline in R?
The text was updated successfully, but these errors were encountered:
When writing data cleaning pipelines, I often find that single step requires more than 80 characters to do right, and to maintain readability, I split the expression over multiple lines. When I start a new pipeline expression, I would like to be able to have the indentation begin back at the original "left-most" point. For example, when I face this problem with dplyr in R, using emacs/ESS, I get indentation that looks like this:
However, the equivalent code in emacs/Julia-mode delivers indentation that looks like this:
Is this intended behavior? Is there a setting I can change to make Julia-mode indent this pipeline the way ESS indents the equivalent pipeline in R?
The text was updated successfully, but these errors were encountered: