I think that to the extent that they don't want a "real" index (and again, I also question that decision), printing the row number makes sense, since that's how you'll access the rows. If I have an array and I select half of its rows, the new array is still indexed 1:n, so they're following the same principle.
It's misleading if you come from an R/python background, but otherwise I can see that it's got its own consistency. On Monday, August 24, 2015 at 7:52:17 PM UTC-6, Robert Smith wrote: > > There was some discussion >> <https://www.google.com/url?q=https%3A%2F%2Fgithub.com%2FJuliaStats%2FDataFrames.jl%2Fissues%2F187&sa=D&sntz=1&usg=AFQjCNGtGVSql7WdSPtRuUaXu6Goner3Wg> >> >> about indexes on Github, but it didn't really get anywhere. I'm also not >> comfortable with that decision to not have indexes. It makes the dataframes >> asymmetric, whereas Pandas' and R's are "matrices with named axes". >> > > Thank you. I'm not sure if that is the most recent conversation about it, > but it has been a while, so maybe we should ask what is the current > consensus on this issue. However, I can see that this is tricky. If there > is a design decision to not rely on indexes for basic functionality, that's > okay. But in the meantime, the solution of having a fake/printed index > looks good generally but provides poor usability. >
