Martin Maechler wrote: > Gregor Gorjanc <gregor.gorjanc <at> bfro.uni-lj.si> writes: > > >>Hello! >> >>Data.frames have new rownames funcionality, however in use of colnames<- >>in R-devel "changes" this. Here is the example: >> >> >>>df1 <- data.frame(letters[1:5]) >>>attributes(df1) >> >>$names >>[1] "letters.1.5." >> >>$row.names >>[1] 1 2 3 4 5 >> >>$class >>[1] "data.frame" >> >> >>>colnames(df1) <- "bla" >>>attributes(df1) >> >>$names >>[1] "bla" >> >>$row.names >>[1] "1" "2" "3" "4" "5" >> >>$class >>[1] "data.frame" >> >>?rownames/colnames help page (R-devel) says that value is coerced to >>character, but why are rownames coerced to character if I assign >>colnames to a data.frame? >> > > > because colnames() and rownames() work via dimnames(), and so do > "colnames<-" and "rownames<-" --- > I think it would be unwise to change this, since it's well documented property > of these functions.
Where? -- the documentation for "colnames<-" on data frames doesn't seem particularly specific about exactly how the replacement will be done. And I don't see any mention of "colnames" in "An Introduction to R". Not changing attr(,"row.names") when colnames are set seems a desirable property. Allowing colnames<- to change attr(,"row.names") damages the nice new feature of data frames that allows them to avoid using a lot of storage on row names. It doesn't look like making "colnames<-" work via "names<-" for dataframes would be particularly contradictory to the documentation. (Also, the documentation is no longer correct for rownames<- for data frames -- it says the value must be character data.) For reference, here's the Details section of ?colnames as it relates to replacement "methods" (these functions don't appear to be generics...?): > The replacement methods for arrays/matrices coerce vector > and factor values of value to character, but do not dispatch > methods for as.character. > > For a data frame, value for rownames should be a character > vector of non-duplicated and non-missing names (this is > enforced), and for colnames a character vector of > (preferably) unique syntactically-valid names. In both > cases, value will be coerced by as.character. (taken from https://svn.r-project.org/R/trunk/src/library/base/man/colnames.Rd) -- Tony Plate > > As you noted yourself, using names() and "names<-" > i.e., > names(df1) <- "bla" > in the example above, is the recommended way for data frames and does not > change > the rownames to character. > > ______________________________________________ > R-devel@r-project.org mailing list > https://stat.ethz.ch/mailman/listinfo/r-devel > ______________________________________________ R-devel@r-project.org mailing list https://stat.ethz.ch/mailman/listinfo/r-devel
