This is exactly the reason I wrote {poorman}. The tidyverse is great for non-programmers. But as soon as something goes wrong and you start looking at the internals, it all goes to 馃挬
This tweet is unavailable
1
1
1
11
I have been saying for a long time that things like the pipe, etc are good for analysis code but should be avoided at nearly all cost for maintained package code. The pipe, and tidyverse more generally optimize a particular thing, and it isn't maintainability/debugability.
1
1
5
Another alternative conclusion, accepting (1) and (2) from yours:
(3) replace current pipeline code with something similar, but simpler (e.g. poorman). Then you don't have to build, or wait/hope for someone else to build better tooling.
1
2
"Forecasting remains difficult, particularly about the future."
Glad I got to demonstrate that again.
Since I said that @_lionelhenry showed possible R parser changes at DSC/ @useR2018_conf and @LukeTierney4 discussed possible implementations at the virtual @useR2020stl
Oct 2, 2020 路 4:13 PM UTC
1
1




